Variable Habitat Selection and Space Use Among Bullsnake (Pituophis Catenifer Sayi) Populations: Distance Between Seasonal Habit

Home , Bullsnake, Pituophis catenifer

Variable habitat selection and space use among bullsnake (Pituophis catenifer sayi)

populations: distance between seasonal habitats drives space use

A Thesis

Submitted to the Faculty of Graduate Studies and Research

In Partial Fulfillment of the Requirements

For the Degree of

Master of Science

Biology

University of Regina

Tera Lynn Edkins

Regina, Saskatchewan

April, 2017

UNIVERSITY OF REGINA

FACULTY OF GRADUATE STUDIES AND RESEARCH

SUPERVISORY AND EXAMINING COMMITTEE

Tera Lynn Edkins, candidate for the degree of Master of Science in Biology, has presented a thesis titled, Variable habitat selection and space use among bullsnake (Pituophis catenifer sayi) populations: distance between seasonal habitats drives space use, in an oral examination held on April 24, 2017. The following committee members have found the thesis acceptable in form and content, and that the candidate demonstrated satisfactory knowledge of the subject material.

External Examiner: *Dr. Donald McAlpine, New Brunswick Museum

Co-Supervisor: Dr. Christopher Somers, Department of Biology

Co-Supervisor: Dr. Raymond Poulin, Department of Biology, Adjunct

Committee Member: Dr. Mark Brigham, Department of Biology

Chair of Defense: Dr. Katya Herman, Faculty of Kinesiology and Health Studies

*Via tele/videoconference

Abstract

The distribution of resources determines space use and habitat selection by snakes. Particularly in northern areas, the proximity of overwintering den sites to summering habitat likely influences space use. The resources driving seasonal movements among habitats may vary among populations and thus, space and habitat requirements may also vary. In addition, human modification may affect resource use by altering available habitats and resources. Although previous studies have reported differences in spatial ecology among populations, the driving factors of this variation remain to be addressed for many species. Bullsnakes (Pituophis catenifer sayi) reach their northern range limits in southern Saskatchewan, where they are currently listed as

Data Deficient by the Committee on the Status of Endangered Wildlife in Canada. Many studies have focused on core range areas, while northern studies have focused on one population in the Frenchman River Valley, Saskatchewan. The habitat and space requirements identified in these previous studies, however, may not be relevant to all bullsnake populations.

I examined bullsnake space and habitat use among independent populations in three river valleys (Frenchman River, South Saskatchewan River, and Big Muddy

Valleys) in Saskatchewan, with varying habitat types (natural and anthropogenic) and availability. I tracked bullsnakes using radio-telemetry, estimated home range areas and movement patterns, and measured third and fourth order habitat selection. The objectives of my research were (1) to examine the extent to which habitat selection and space use may vary among populations and (2) to identify important habitat features in common among snakes from different populations.

Saskatchewan bullsnakes demonstrated variable space use and movement patterns among populations, exceeding space use requirements previously observed in southern areas. One population (Big Muddy Valley), on average, used 2.7 to 3 times less space, travelled 2.3 to 2.7 times shorter distances from overwintering sites, and had greater home range overlap than snakes from the other populations (Frenchman and South

Saskatchewan River Valleys). This suggests that bullsnakes in the Big Muddy Valley had a closer spatial association between seasonal habitats.

Bullsnakes appear to be flexible in terms of their third order habitat selection.

Native habitats were used as expected across all valleys, but human-modified habitats were used at different frequencies across populations. These differences in habitat selection among populations are most likely due to differences in habitat availability among landscapes. Fourth habitat selection, however, was similar among populations, with bullsnakes selecting for sites typically within 1 m of a refuge site (including burrows, cement pads, and rock piles).

It appears that bullsnakes occupy variable-sized home ranges and move variable distances. Bullsnakes also appear to be flexible in terms of how they meet resource requirements across their geographic range via habitat selection at a broad spatial scale.

My study did find, however, that at a fine spatial scale refuge sites are an important habitat feature for bullsnakes. Conservation and management strategies are typically broad and are implemented as if populations of the same species have similar habitat requirements. However, my results indicate that this is not the case. As such, management plans may not be applicable to all populations.

III

Acknowledgements

First I would like to thank my co-supervisors Dr. Chris Somers and Dr. Ray

Poulin for their support. Thank you for putting your faith in a snake newbie and for allowing me to gain experience in the field of conservation biology. Thank you to my committee member Mark Brigham for his valuable input and support over the course of my thesis. Thank you to Dr. Mark Vanderwel for his statistical guidance.

I am grateful to my wonderful field assistants Leagh Vermeylen, Ana Pecorari, and Allie Gallon for the long hours they put in tracking snakes. Thank you for being patient with me as we learned the ropes of radio-telemetry together. Thanks to Danae

Frier for teaching me the field techniques I required to complete my project. Thank you to my lab mates at the University of Regina for your helpful advice and for being the friendliest and most supportive lab group a grad student could ask for.

I thank the Saskatchewan Parks staff and Big Muddy Valley landowners for their support and cooperation during the field season. Thank you to Dr. Tracy Fisher and the

North Albert Veterinary Clinic for completing our snake surgeries for summer 2015. I also thank Dr. Miranda Sidar and Dr. Dennilyn Parker, in addition to the veterinarians of the Western College of Veterinary Medicine, for donating their services for the snake surgeries in 2016.

I would like to thank the Friends of the Royal Saskatchewan Museum for presenting me with the first ever RSM Graduate Student Scholarship. I would also like to thank The Faculty of Graduate Studies and Research for the financial support they provided. Finally, I would like to thank the Royal Saskatchewan Museum for funding this research project.

Dedication

I would like to dedicate this thesis to my mom, dad and sister. I cannot describe how much I appreciate everything you have done for me. Thank you for your constant encouragement and for always supporting my dreams, even if they were to be a crazy snake-studying scientist. I would also like to dedicate this thesis to my best friend and partner, Daniel Blerot. This thesis would not exist if you had not supported me when I decided to move to Saskatchewan to get my Master’s degree. You always believe in me. I am incredibly grateful for everything you have done for me and everything you have put up with throughout this process. I love you and am looking forward to spending my life with you.

Table of contents

Abstract…………………………………………………………………………………...II

Acknowledgements………………………………………………………………………IV

Dedication………………………………………………………………………………...V

Table of contents…………………………………………………………………………VI

List of tables……………………………………………………………………………VIII

List of figures……………………………………………………………………...... XIII

Chapter 1: General Introduction……………………………………………………...... 1

Habitat selection by animals………………………………………………………2

Animal space use………………………………………………………………….4

Habitat selection in human-modified areas…………………………………...... 5

Habitat selection in reptiles………………………………………………………..6

Habitat selection by snakes…………………………………………………...... 7

Snakes in Saskatchewan…………………………………………………………..8

References…………………………………………………………………...... 11

Chapter 2: Variable habitat selection and space use among bullsnake (Pituophis catenifer sayi) populations: distance between seasonal habitats drives space use…...... 24

Introduction………………………………………………………………………25

Materials and Methods…………………………………………………………...28

Results……………………………………………………………………………38

Discussion…………………………………………………………………...... 51

Conclusions………………………………………………………………………61

References…………………………………………………………………...... 63

Chapter 3: Synthesis……………………………………………………………………..73

Conservation and management implications…………………………………….74

Future research…………………………………………………………………...78

References…………………………………………………………………...... 82

VII

List of Tables

Chapter 2:

Table 1. Mean (±SD) telemetry data, maximum displacement, and home range overlap of bullsnakes from three different river valleys in southern Saskatchewan: the Big Muddy

Valley (BMV), South Saskatchewan River Valley (SSRV), and the Frenchman River

Valley (FRV). N = total number of snakes tracked..…………………………………….39

(AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights)...……..41

VIII

standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights).………………………………43

Table 4. Top generalized linear model, null model, and all models with ΔAIC <3 evaluating the best predictor of Saskatchewan bullsnake 95% KDE home range area size.

Fixed effects included river valley, distance to nearest man-made structure (m), snout to vent length (SVL, cm), and snake sex. The number of model parameters (K), Akaike’s

Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights).………………………………44

Table 6. Regression analyses evaluating the differences in home range overlap between the Frenchman River (FRV), Big Muddy (BMV), and South Saskatchewan River (SSRV)

Valleys. Home range overlap was measured in two ways: as the distance between minimum convex polygon centroids and the proportion of home range overlap between snakes. A gamma regression was used to examine the effect of river valley on centroid distance. A beta regression was used examine the effect of river valley on proportion of home range overlap. The fixed effect for both models was river valley. Presented here are the parameter estimates, standard errors (SE), t-values/z-values, and p-values..………..48

Table 8. Top generalized linear mixed model, null model, and all models with Delta AIC

<2 used for evaluating bullsnake habitat selection in the South Saskatchewan River

Valley. Selection was examined using radio-telemetry locations compared to available

locations. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), maximum vegetation height (cm), and

Robel pole vegetation density measurements. Random effect was individual snake ID.

Presented here are the number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model

(ΔAIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the

Akaike weights).………...……………………………………………………………….53

Table 9. Top generalized linear mixed model, null model, and all models with Delta AIC

<2 used for evaluating bullsnake habitat selection at the local scale. Selection was examined using measurements taken 10 m in each cardinal direction at used and available sites in the Big Muddy Valley. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), Robel pole vegetation density measurements, and maximum vegetation height (cm). Random effect was individual snake ID. Presented here are the number of parameters (K), Akaike’s

Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (Delta AIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights)…...... 55

Table 10. Top generalized linear mixed model, null model, and all models with Delta

AIC <2 used for evaluating bullsnake habitat selection at the local scale. Selection was examined using habitat measurements taken 10 m in each cardinal direction at used and available sites in the South Saskatchewan River Valley. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), Robel pole vegetation density measurements, and maximum vegetation height

(cm). Random effect was individual snake ID. Presented here are the number of parameters (K), Akaike’s Information Criterion value corrected for small sample size

(AICc), difference in AICc from the top model (Delta AIC), and the Akaike weights.

Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals

(CI), and the importance values (calculated using the Akaike weights)..…………...…...56

XII

List of Figures

Chapter 2:

Figure 1. Study locations where bullsnakes were tracked using radio-telemetry in southern Saskatchewan: the Frenchman River Valley (FRV, 2008-2009, data collected by

Martino et al. 2012), Big Muddy Valley (BMV, 2015), and South Saskatchewan River

Valley (SSRV, 2016), indicated by the black pentagons. The North American bullsnake range is highlighted in dark grey..……………………………………………………….30

Figure 2. Minimum convex polygons (MCPs) for bullsnakes in the Frenchman River

(FRV), South Saskatchewan River (SSRV), and Big Muddy (BMV) Valleys. MCPs are shown for 9 Frenchman River Valley, 8 South Saskatchewan River Valley, and 4 Big

Muddy Valley bullsnakes. Den sites are indicated by stars, roads by thick black lines, urban areas by dark grey polygons, crop fields by crosshatched polygons, and lakes by light grey polygons. Scale differs among panels…………………………………….…..42

Figure 3. Box and Whisker Plot of home ranges for bullsnakes in the Frenchman River

Valley (FRV; N=14), Big Muddy Valley (BMV; N=7), and South Saskatchewan River

Valley (SSRV; N=14). The median is displayed as the thick black bar; boxes represent the first and third quartiles; whiskers show the highest and lowest values within 1.5 times the interquartile range. Dots indicate outliers. The 100% Minimum Convex Polygon

(MCP) is shown in dark grey, the 95% Kernel Density Estimate in light grey, and the

50% Kernel Density Estimate in white...………………………………………..……….46

XIII

Figure 4. Percent of different habitats used and available within a 1.3 km buffer surrounding den sites in the Big Muddy Valley, Saskatchewan.………………………...49

Figure 5. Percent of different habitat types used and available within a 2.4 km buffer surrounding den sites in the South Saskatchewan River Valley within Saskatchewan

Landing Provincial Park, Saskatchewan.………………………………………………...50

Figure 6. Relative frequency of habitat use by bullsnakes in the Big Muddy Valley

(BMV) and South Saskatchewan River Valley (SSRV) at locations 0 to 1 m, 2 to 5 m, and greater than 6 m from a burrow..…………………………………………………….54

XIV

Chapter 1

General Introduction

Habitat selection by animals

Habitat selection is a behavioral choice made by animals, connecting individuals to required resources (Buskirk and Millspaugh 2006). Habitats are selected based on various abiotic and biotic factors, such as environmental conditions (Hill 2006), predator avoidance (Mao et al. 2005; Hill 2006; Wisler et al. 2008), vegetation structure (Bergin

1992; Anich et al. 2012; Danielsen et al. 2014), as well as the availability and distribution of resources, such as food (Korte 2008; Ahlers et al. 2010), refuge sites (Aguilar and

Cruz 2010; Grillet et al. 2010), and nesting sites (Blaum et al. 2007). Factors driving habitat selection vary among species and populations (Benson and Chamberlain 2007), as well as within populations, where animals of different ages, sexes (Waldron et al. 2006;

Ranglack and du Toit 2015), reproductive states (Harvey and Weatherhead 2006), and body sizes (Milne and Bull 2000) may use habitat differently. In addition, habitat selection is usually driven by multiple factors acting simultaneously (Hill 2006; LaForge et al. 2016). For example, wood turtles (Clemmys insculpta) select and move between different habitats to satisfy both foraging and thermoregulatory needs (Compton et al.

2002). As such, to fully understand animal habitat selection we must understand the driving factors.

Habitat selection occurs at different spatial scales, typically ranging from first to fourth order (Johnson 1980). First and second order selection describe the geographic range of a species and the home range of an individual (Johnson 1980). Third and fourth order selection describe broad and local scale habitat selection within a home range

(Johnson 1980). Habitat selection at one scale does not necessarily predict habitat selection at another (Bergin 1992; Mayor et al. 2009). For example, yellow mongooses

(Cynictis penicillata) select primarily for shrubby areas at a fine scale, as shrubs provide

refuge and nesting sites (Blaum et al. 2007). However, at a broader spatial scale, mongooses select primarily for open areas, where prey occur (Blaum et al. 2007). As such, Bergin (1992) has recommended that studies examine habitat selection at multiple spatial scales.

Radio-telemetry is one of the most common field techniques used to assess whether individuals of a population use habitats randomly, to rank habitats based on their relative use, and to relate use to resource availability and abundance (Aebischer et al.

1993). As selected habitats are a subset of available ones (Buskirk and Millspaugh 2006), habitats are considered to be selected if they are used disproportionately more than they are available (Johnson 1980; Aebischer et al. 1993; Mayor et al. 2009). Designation of habitat availability itself depends on the accessibility of a habitat to the species of interest

(Johnson 1980). Animals may not make use of all available habitats and not all habitats in an area may be considered available to all populations of the same species (Aebischer et al. 1993; Benson and Chamberlain 2007). As a result, individual and population habitat selection will depend, in part, on the habitats that are considered to be available (Johnson

1980). For analytical reasons, available habitat types should be designated prior to data collection (Buskirk and Millspaugh 2006).

Using telemetry data, compositional analysis permits the ranking of habitats based on their use and individual preference (Johnson 1980; Aebischer et al. 1993). The terms selection and preference are commonly interchanged, though preference is defined separately from selection and is based on the likelihood of a habitat being chosen if all habitats are equally available (Johnson 1980). This is typically not the case in the wild, where habitats are rarely equally available. As such, employing a relative scale of habitat

selection, from least to most occupied, allows for inferred preference based on the use of one habitat relative to others (Aebischer et al. 1993).

Animal space use

Resources are spatially distributed across a landscape and animals move through space to acquire the resources that allow them to meet the necessities of life (Boggie and

Mannan 2014; Moorter et al. 2016). Home range is a measure of animal space use and is defined as the area where regular activities, such as feeding, mating, and reproduction, are carried out (Burt 1943; Aebischer et al. 1993). The most common measures of home range are the minimum convex polygon (MCP, 100%) and kernel density estimates

(KDE, 50% and 95%). Both measures have their merits and drawbacks. For example,

MCP estimates give equal weight to all organism relocations, while KDEs allow for the examination of the intensity of space use in different areas of the home range (Mitchell and Powell 2008). However, KDEs depend on a selected smoothing factor, which can drastically affect the total area of the estimate (Mitchell and Powell 2008). As such, it is important to use the optimal home range estimator and measure(s) of space use for the research questions of interest.

Resource abundance, quality, and availability ultimately determine animal space use requirements. When resources are abundant, animals require less space to obtain them and thus, can exist in smaller home ranges (Okarma et al. 1998; Mattison et al.

2013). This is also the case when resource quality is high (Whitaker et al. 2007; Korte

2008; Mattisson et al. 2013). Areas with limited resources, however, can require a greater use of space to meet their requirements. For example, high prey abundance in an area results in snakes using smaller home ranges (Ettling et al. 2016). Conversely, snakes travel longer distances and inhabit larger home ranges when prey are sparsely distributed

(Duvall et al. 1990). In terms of developing successful conservation and management plans, it is important to understand how organisms use space and which resources are driving space use (Anich et al. 2012; Mattisson et al. 2013).

Habitat selection in human-modified areas

Human modification of the natural environment affects landscape configurations by introducing anthropogenic habitat types, with varying resource availability and quality. As a result, human land use has variable effects on the habitat and space use patterns of the animals occupying these modified landscapes. Frequently, these effects are negative, with many organisms modifying their spatial ecology to avoid areas of human land use, such as settlements, roads, and crop lands (Juang et al. 2007; Larsen and

Guillemette 2007; Johnson and Russell 2014). However, animals may also select for modified habitats as features associated with these habitats may provide higher quality resources (Bellamy et al. 2000). For example, Cooper’s hawks (Accipiter cooperii) within a large urban center demonstrated higher relative densities than natural areas, due to higher prey abundances in the urban habitat (Boggie and Mannan 2014).

Modified habitats can also introduce an increased risk of mortality via human- wildlife conflicts (Mattisson et al. 2013; Bonnot et al. 2013; Belton et al. 2016). As such, behavioral modifications may occur to avoid the risks associated with human land use.

For example, large mammals, such as roe deer (Capreolus capreolus) and African elephants (Loxodonta africana), make use of modified habitats at night to feed, while avoiding these habitats during the day in favor of habitats with greater cover (Graham et al. 2009; Bonnot et al. 2013). Overall, the response of a population to human land use will vary depending on the species-specific resource requirements and the quality and abundance of available resources (Driscoll 2004; Corey and Doody 2010).

Habitat selection in reptiles

Reptiles are an ectothermic group of organisms that manage their body temperature via heat transfer with the external environment (Dzialowski et al. 2005). As drastic increases or decreases in body temperature can effectively reduce fitness and may increase mortality (Huey and Kingsolver 1993; Blouin-Demers and Nadeau 2005;

Blouin-Demers and Weatherhead 2008; Aragon et al. 2010), this group must behaviorally regulate their body temperature to avoid temperature extremes (Huey et al. 2012).

Endotherms, which include birds and mammals, differ from ectotherms as they primarily produce their own body heat. Most endotherms are also homeothermic, meaning that they use behavioral, morphological, and physiological modifications to maintain a high and relatively constant body temperature under various environmental conditions (Scholander

1955; Huey et al. 2012).

Reptiles are currently an understudied group, with many species lacking basic biological and population data (Seburn and Seburn 2000; Lesbarreres et al. 2014;

Griffiths et al. 2015). One reason for this data deficiency is the fact that many reptiles are difficult to detect in their natural environment (Mazerolle et al. 2007; Griffiths et al.

2015). Many species are nocturnal, have cryptic coloration, and are active only during certain times of year or under certain environmental conditions (Seburn and Seburn 2000;

Mazerolle et al. 2007; Griffiths et al. 2015). In addition, for many smaller-bodied reptile species, radio-transmitters are too large to be implanted or attached externally. These factors have made it difficult to collect various biological data and to make confident estimates of population size.

Currently in Canada, where many reptiles reach their northern range limits, 77% of species are considered to be at risk of extinction (COSEWIC; Lesbarreres et al. 2014),

with primary conservation threats including road mortality and habitat loss/fragmentation via agriculture, energy production, mining, forestry, and water management (Seburn and

Seburn 2000; Lesbarreres et al. 2014). However, little relevant information is currently available to inform conservation decisions for many reptile taxa in this region

(Lesbarreres et al. 2014; Griffiths et al. 2015). In addition, reptile populations at northern latitudes are typically sparse and must deal with more difficult environmental conditions than may be present farther south. As a result, threats to reptile populations at northern range limits are exacerbated. To make informed conservation and management decisions, we need to better understand habitat use and space requirements, and to identify other factors important for Canadian reptiles.

Habitat selection by snakes

Globally, snake populations are in decline, mainly due to habitat loss via agriculture, natural resource use, and urban development (Reading et al. 2010; Bohm et al. 2013). About 12% of snake species are threatened worldwide, with 24% being data deficient, meaning we lack the appropriate data to assess their conservation status (Bohm et al. 2013). Compared to many other reptile groups, snakes also face human persecution, as they tend to be highly feared and disliked (Seburn and Seburn 2000). Many snakes, venomous species in particular, are actively killed by humans due to their perceived threat (Seburn and Seburn 2000). Traffic mortality is also high for snakes, as many snakes use roads for thermoregulatory purposes. As a result snakes on roads are at an increased risk of snake-vehicle collisions, whether accidental or on purpose (Seburn and

Seburn 2000; Fortney et al. 2012).

Presently, there are an increasing number of studies examining snake habitat selection and space use patterns for a diversity of species (Johnson et al. 2007; Carfagno

and Weatherhead 2008; Steen et al. 2010; Bailey et al. 2012; Shipley et al. 2013;

Zappalorti et al. 2015). However habitat and space use have typically only been addressed for one or a few populations of a species. Studies have also demonstrated that space use varies among snake populations regardless of whether they occupy natural or human-modified areas (Corey and Doody 2010; Anguiano and Diffendorfer 2015; Smith et al. 2015; Ettling et al. 2016). However, these studies addressed snake habitat use exclusively in one landscape configuration or set of available habitats. As such, there is a need to increase the breadth of studies in terms of identifying important snake habitat and spatial requirements, and how these requirements may differ among populations.

Snakes in Saskatchewan

In Canada, habitat loss is the primary threat to snake populations (Lesbarreres et al. 2014); specifically the disconnection between critical habitats (such as summering and overwintering habitats) created by human modification (Lesbarreres et al. 2014). This is especially the case at higher latitudes where overwintering den sites are potentially a limiting resource (Burger et al. 1988; Harvey and Weatherhead 2006; Gardiner et al.

2013). Populations also tend to be less dense at northern latitudes, which exacerbates man-made threats. Many of the snake species occupying the grasslands of Manitoba,

Saskatchewan, and Alberta may be at risk, but there is insufficient information for proper assessment of conservation and management strategies (Lesbarreres et al. 2014).

Unfortunately, studies regarding habitat selection and space use by grassland snakes are relatively sparse (Gannon and Secoy 1985; Martino et al. 2012; Gardiner et al. 2013), with the majority of Canadian-snake studies focusing on non-grassland snake communities (Blouin-Demers and Weatherhead 2001a; 2001b; Harvey and Weatherhead

2006; Williams et al. 2012; Gomez et al. 2015).

Southern Saskatchewan is home to a diversity of snake species, many reaching their northern range limits in this area (Seburn and Seburn 2000). Previous studies have addressed the habitat and space use of three species in the Frenchman River Valley: the eastern yellow-bellied racer (Coluber constrictor flaviventris), prairie rattlesnake

(Crotalus viridis viridis), and bullsnake (Pituophis catenifer sayi). These studies found that snakes make long distance migrations between summer habitat and den sites and that habitat selection in southern Saskatchewan differs among species as well as from populations in the United States (Martino et al. 2012; Gardiner et al. 2013). These authors also concluded that current management strategies are insufficient to protect snakes based on known habitat and space use patterns. For example, the eastern yellow-bellied racer would not be protected within the 500 m conservation buffer currently provided around overwintering den sites, as this species moves away from the denning area in the summer

(Parks Canada 2010; Martino et al. 2012; Gardiner et al. 2013). As habitat and space use patterns can vary among populations, it is necessary to determine if the findings from the

Frenchman River Valley are relevant to populations across the Canadian range of these species.

Bullsnakes are the largest snake species in Canada (Wright 2016). However, the spatial ecology of this species at northern latitudes, where environmental conditions are harsher and resources may be more limited or harder to obtain, remains to be thoroughly examined. Studies focusing on populations in core range areas show that bullsnake space use and habitat selection may be quite variable among populations (Moriarty and Linck

1997; Rodriguez-Robles 2003; Kapfer et al. 2008; 2010). Bullsnakes are currently listed as Data Deficient (COSEWIC 2002), meaning there is insufficient data to assess the conservation status of bullsnake populations in Canada. I hypothesized that the spatial

distribution of resources in the environment, and thus snake space and habitat use, are influenced by habitat availability and landscape configuration. As such, I predicted that space and habitat use would vary among bullsnake populations. I suggest that this is due to differences in available habitat types as well as landscape configurations and modifications among river valleys. The specific objectives of my study were:

1. To measure the variation in space use and habitat selection among different

bullsnake populations in Saskatchewan

2. To identify important habitat features in common among bullsnake populations

in Saskatchewan.

References

Aebischer, N. J., Robertson, P. A., and Kenward, R. E. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology, 74(5): 1313-1325.

Aguilar, R. and Cruz, F. B. 2010. Refuge use in a Patagonian nocturnal lizard, Homonota darwini: the role of temperature. J. Herpetol. 44(2): 236-241.

Ahlers, A. A., Heske, E. J., Schooley, R. L., and Mitchell, M. A. 2010. Home ranges and space use of muskrats Ondatra zibethicus in restricted linear habitats. Wildl. Biol. 16(4):

400-408.

Anguiano, M. P. and Diffendorfer, J. E. 2015. Effects of fragmentation on the spatial ecology of the California kingsnake (Lampropeltis californiae). J. Herpetol. 49(3):

420-427.

Anich, N. M., Benson, T. J., and Bednarzi, J. C. 2012. What factors explain differential use within Swainson’s warbler (Limnothlypis swainsonii) home ranges? The Auk, 129(3):

409-418.

Aragon, P., Lobo, J. M., Olalla-Tarraga, M. A., and Rodriguez, M. A. 2010. The contribution of contemporary climate to ectothermic and endothermic vertebrate distributions in a glacial refuge. Global Ecol. Biogeogr. 19: 40-49.

Bailey, R. L., Campa III, H., Bissell, K. M., and Harrison, T. M. 2012. Resource selection by the eastern massasauga rattlesnake on managed land in Southwestern Michigan. J.

Wildl. Manage. 76(2): 414-421.

Bellamy, P. E., Shore, R. F., Ardeshir, D., Treweek, J. R., and Sparks, T. H. 2000. Road verges as habitat for small mammals in Britain. Mammal Rev. 30(2): 131-139.

Belton, L. E., Cameron, E. Z., and Dalerum, F. 2016. Spotted hyaena space use in relation to human infrastructure inside a protected area. PeerJ, 4: e2596.

Benson, J. F. and Chamberlain, M. J. 2007. Space use and habitat selection by female

Louisiana black bears in the Tensas River Basin of Louisiana. J. Wildl. Manage. 71(1):

117-126.

Bergin, T. M. 1992. Habitat Selection by the western kingbird in western Nebraska: a hierarchical analysis. The Condor, 94(4): 903-911.

Blaum, N., Rossmanith, E., Fleissner, G., and Jeltsch, F. 2007. The conflicting importance of shrubby landscape structures for the reproductive success of the yellow mongoose (Cynictis penicillata). J. Mammal. 88(1): 194-200.

Blouin-Demers, G. and Weatherhead, P. J. 2001a. An experimental test of the link between foraging, habitat selection and thermoregulation in black rat snakes Elaphe obsoleta obsolete. J. Anim. Ecol. 70: 1006-1013.

Blouin-Demers, G. and Weatherhead, P. J. 2001b. Habitat use by black rat snake (Elaphe obsoleta obsoleta) in fragmented forests. Ecology, 82(10): 2882-2896.

Blouin-Demers, G. and Nadeau, P. 2005. The cost-benefit model of thermoregulation does not predict lizard thermoregulatory behavior. Ecology, 86(3): 560-566.

Blouin-Demers, G. and Weatherhead, P. J. 2008. Habitat use is linked to components of fitness through the temperature-dependence of performance in ratsnakes (Elaphe obsoleta). Is. J. Ecol. Evol. 54: 361-372.

Boggie, M. A. and Mannan, R. W. 2014. Examining seasonal patterns of space use to gauge how an accipiter responds to urbanization. Landsc. Urban Plann. 124: 34-42.

Bohm, M., Collen, B., Baillie, J. E. M., Bowler, P., Chanson, J., et al. 2013. The conservation status of the world’s reptiles. Biol. Conserv. 157: 372-385.

Bonnot, N., Morellet, N., Verheyden, H., Cargnelutti, B., Lourtet, B., Klein, F., and

Hewison, A. J. M. 2013. Habitat use under predation risk: hunting, roads and human dwellings influence the spatial behavior of roe deer. Eur. J. Wildl. Res. 59: 185-193.

Burger, J., Zappalorti, R. T., Gochfeld, M., Boarman, W. I., Caffrey, M., Doig, V.,

Garber, S. D., Lauro, B., Mikovsky, M., Safina, C., and Saliva, J. 1988. Hibernacula and

summer den sites of pine snakes (Pituophis melanoleucus) in the New Jersey pine barrens. J. Herpetol. 22(4): 425-433.

Burt, W. H. 1943. Territoriality and home range concepts as applied to mammals. J.

Mammal. 24(3): 346-352.

Buskirk, S. W. and Millspaugh, J. J. 2006. Metrics for studies of resource selection. J.

Wildl. Manage. 70(2): 358-366.

Carfagno, G. L. F. and Weatherhead, P. J. 2008. Energetics and space use: intraspecific and interspecific comparisons of movements and home ranges of two Colubrid snakes. J.

Anim. Ecol. 77: 416-424.

Committee on the Status of Endangered Wildlife in Canada (COSEWIC). 2002.

COSEWIC assessment and status report on the Gophersnake Pituophis catenifer in

Canada. Ottawa. pp. 1- 33.

Compton, B. W., Rhymer, J. M., and McCollough, M. 2002. Habitat selection by wood turtles (Clemmys insculpta): an application of paired logistic regression. Ecology, 83(3):

833-843.

Corey, B. and Doody, J. S. 2010. Anthropogenic influences on the spatial ecology of a semi-arid python. J. Zool. 281: 293-302.

Danielsen, A. K., Rutherford, P., and Koper, N. 2014. The importance of vegetation structure and artificial cover for prairie skinks (Plestiodon septentrionalis) on exurban land. Herpetologica, 48(1): 67-73.

Driscoll, D. A. 2004. Extinction and outbreaks accompany fragmentation of a reptile community. Ecol. Appl. 14(1): 220-240.

Duvall, D., Goode, M. J., Hayes, W. K., Leonhardt, J. K., and Brown, D. G. 1990. Prairie rattlesnake vernal migration: field experimental analyses and survival value. Nat. Geog.

Res. 6(4): 457-469.

Dzialowski, E. M. 2005. Use of operative temperature and standard operative temperature models in thermal biology. J. Therm. Biol. 30: 317-334.

Ettling, J. A., Aghasyan, L. A., Aghasyan, A. L., and Parker, P. G. 2016. Spatial ecology of Armenian vipers, Montivipera raddei, in two different landscapes: human-modified vs. recovered natural. Russ. J. Herpetol. 23(2): 93-102.

Fortney, A. N., Poulin, R. G., Martino, J. A., Parker, D. L., and Somers, C. M. 2012.

Proximity to hibernacula and road type influence potential road mortality in snakes in southwestern Saskatchewan. Can. Field Nat. 126: 194-203.

Gannon, V. P. J. and Secoy, D. M. 1984. Seasonal and daily activity patterns in a

Canadian population of the prairie rattlesnake, Crotalus viridus viridis. Can. J. Zool.

63: 86-91.

Gardiner, L. E., Somers, C. M., Martino, J. A., Parker, D. L., and Poulin, R. G. 2013.

Balancing the dumbbell: summer habitats need protection in addition to winter dens for northern snake communities. J. Wildl. Manage. 77(5): 975-982.

Gomez, L., Larsen, K. W., and Gregory, P. T. 2015. Contrasting patterns of migration and habitat use in neighboring rattlesnake populations. J. Herpetol. 49(3): 371-376.

Graham, M. D., Douglas-Hamilton, I., Adams, W. M., and Lee, P. C. 2009. The movement of African elephants in a human-dominated land-use mosaic. Anim. Conserv.

12: 445-455.

Griffiths, R. A., Foster, J., Wilkinson, J. W., and Sewell, D. 2015. Science, statistics and surveys: a herpetological perspective. J. Appl. Ecol. 52: 1413-1417.

Grillet, P., Cheylan, M., Thirion, J-M., Dore, F., Bonnet, X., Dauge, C., Chollet, S., and

Marchand, M. A. 2010. Rabbit burrows or artificial refuges are a critical habitat component for the threatened lizard, Timon lepidus (Sauria, Lacertidae). Biodivers.

Conserv. 19: 2039-2051.

Harvey, D. S. and Weatherhead, P. J. 2006. A test of the hierarchical model of habitat

selection using eastern massasauga rattlesnakes (Sistrurus c. catenatus). Biol. Conserv.

130: 206-216.

Hill, R. A. 2006. Thermal constraints on activity scheduling and habitat choice in baboons. Am. J. Phys. Anthropol. 129: 242-249.

Huey, R. B. and Kingsolver, J. G. 1993. Evolution of resistance to high temperature in ectotherms. Am. Nat. 142: S21-S46.

Huey, R. B., Kearney, M. R., Krockenberger, A., Holtum, J. A. M., Jess, M., and

Williams, S. E. 2012. Predicting organismal vulnerability to climate warming: roles of behaviour, physiology and adaptation. Phil. Trans. R. Soc. B. 367: 1665-1679.

Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology, 61(1): 65-71.

Johnson, C. J. and Russell, D. E. 2014. Long-term distribution responses of a migratory caribou herd to human disturbance. Biol. Conserv. 177: 52-63.

Johnson, R. W., Fleet, R. R., Keck, M. B., and Rudolph, D. C. 2007. Spatial ecology of the coachwhip, Masticophis flagellum (Squamata: Colubridae), in eastern Texas.

Southeast Nat. 6(1): 111-124.

Juang, G., Zhang, M., and Jianzhang, M. A. 2007. Effects of human disturbance on movement, foraging and bed selection in red deer Cervus elaphus xanthopygus from the

Wandashan Mountains, northeastern China. Acta Theriol. 52(4): 435-446.

Kapfer, J. M., Coggins, J. R., and Hay, R. 2008. Spatial ecology and habitat selection of bullsnakes (Pituophis catenifer sayi) at the northern periphery of their geographic range.

Copeia, 2008(4): 815-826.

Kapfer, J. M., Pekar, C. W., Reineke, D. M., Coggins, J. R., and Hay. R. 2010. Modeling the relationship between habitat preferences and home-range size: a case study on a large mobile Colubrid snake from North America. J. Zool. 282: 13-20.

Korte, L. M. 2008. Habitat selection at two spatial scales and diurnal activity patterns of adult female forest buffalo. J. Mammal. 89(1): 115-125.

LaForge, M. P., Michel, N. L., Wheeler, A. L., and Brook, R. K. 2016. Habitat selection by female moose in the Canadian prairie ecozone. J. Wildl. Manage. 80(6): 1059-1068.

Larsen, J. K. and Guillemette, M. 2007. Effects of wind turbines on flight behaviour of wintering common eiders: implications for habitat use and collision risk. J. Appl. Ecol.

44: 516-522.

Lesbarreres, D., Ashpole, S. L., Bishop, C. A., Blouin-Demers, G., Brooks, R. J.,

Echaubard, P., Govindarajulu, P., Green, D. M., Hecnar, S. J., Herman, T., Houlahan, J.,

Litzgus, J. D., Mazerolle, M. J., Paszkowski, C. A., Rutherford, P., Schock, D. M.,

Storey, K. B., and Lougheed, S. C. 2014. Conservation of herpetofauna in northern landscapes: threats and challenges from a Canadian perspective. Biol. Conserv. 170: 48-

55.

Mao, J. S., Boyce, M. S., Smith, D. W., Singer, F. J., Vales, D. J., Vore, J. M., and

Merrill, E. H. 2005. Habitat selection by elk before and after wolf reintroduction in

Yellowstone National Park. J. Wildl. Manage. 69(4): 1691-1707.

Martino, J. A., Poulin, R. G., Parker, D. L., and Somers, C. M. 2012. Habitat selection by grassland snakes at northern range limits: implications for conservation. J. Wildl.

Manage. 76(4): 756-767.

Mattison, J., Sand, H., Wabakken, P., Gervasi, V., Liberg, O., Linnell, J. D. C., Rauset,

G. R., and Pedersen, H. C. 2013. Home range size variation in a recovering wolf population: evaluating the effect of environmental, demographic, and social factors.

Oecologia, 173: 813-825.

Mayor, S. J., Schneider, D. C., Schaefer, J. A., and Mahoney, S. P. 2009. Habitat selection at multiple scales. Ecoscience, 16(2): 238-247.

Mazerolle, M. J., Bailey, L. L., Kendall, W. L., Royle, J. A., Converse, S. J., and Nichols,

J. D. 2007. Making great leaps forward: accounting for detectability in herpetological field studies. J. Herpetol. 41(4): 672-689.

Milne, T. and Bull, C. M. 2000. Burrow choice by individuals of different sizes in the endangered pygmy blue tongue lizard Tiliqua adelaidensis. Biol. Conserv. 95: 295-301.

Mitchell, M. S. and Powell, R. A. 2004. A mechanistic home range model for optimal use of spatially distributed resources. Ecol. Modell. 177: 209-232.

Moorter, B. V., Rolandsen, C. M., Basille, M., and Gaillard, J-M. 2016. Movement is the glue connecting home ranges and habitat selection. J. Anim. Ecol. 85: 21-31.

Moriarty, J. J. and Linck, M. 1997. Reintroduction of bullsnakes into a recreated prairie.

In: Moriarty, J. J. and Jones, D. 1997. Minnesota’s amphibians and reptiles: their conservation and status. Serpent’s Tale Natural History Book Distributors, Lanesboro,

Minnesota. pp. 43-54.

Okarma, H., Jedrzejewski, W., Schmidt, K., Sniezko, S., Bunevich, A. N., and

Jedrzejewska, B. 1998. Homes ranges of wolves in Bialowieza Primeval Forest, Poland, compared with other Eurasian populations. J. Mammal. 79(3): 842-852.

Parks Canada Agency. 2010. Recovery strategy for eastern yellow-bellied racer (Coluber constrictor flaviventris) in Canada. Species at Risk Act Recovery Strategy Series. Parks

Canada Agency. Ottawa. p. 12.

Ranglack, D. H. and du Toit, J. T. 2015. Habitat selection by free-ranging bison in a

mixed grazing system on public land. Rangeland Ecol. Manage. 68: 349-353.

Rodriguez-Robles, J. A. 2003. Home ranges of gopher snakes (Pituophis catenifer,

Colubridae) in central California. Copeia, 2003(2): 391-396.

Reading, C. J., Luiselli, L. M., Akani, G. C., Bonnet, X., Amori, G., Ballouard, J. M.,

Filippi, E., Naulleau, G., Pearson, D., and Rugiero, L. 2010. Are snake populations in widespread decline? Biol. Lett. 6: 777-780.

Russell, D. E. and Johnson, C. J. 2014. Long-term distribution responses of a migratory caribou herd to human disturbance. Biol. Conserv. 177: 52-63.

Scholander, P. F. 1955. Evolution of climatic adaptation in homeotherms. Evolution,

9(1): 15-26.

Seburn, D. and Seburn, C. 2000. Conservation priorities for the amphibians and reptiles of Canada. World Wildlife Fund. Canadian Amphibian and Reptile Conservation

Network.

Shipley, B. K., Chiszar, D., Fitzgerald, K. T., and Saviola, A. J. 2013. Spatial ecology of prairie rattlesnakes (Crotalus viridis) associated with black-tailed prairie dog (Cynomys ludovicianus) colonies in Colorado. Herpetol. Conserv. Biol. 8(1): 240-250.

Smith, R. M., Spotila, J. R., and Bien, W. F. 2015. Spatial ecology of northern pinesnakes

(Pituophis m. melanoleucus) in disturbed and undisturbed habitats in the New Jersey pine barrens. Herpetologica, 71(1): 19-25.

Steen, D. A., Linehan, J. M., and Smith, L. L. 2010. Multiscale habitat selection and refuge use of common kingsnakes, Lampropeltis getula, in Southwestern Georgia.

Copeia, 2010(2): 227-231.

Waldron, J. L., Lanham, J. D., and Bennett, S. H. 2006. Using behaviorally-based seasons to investigate canebrake rattlesnake (Crotalus horridus) movement patterns and habitat selection. Herpetologica, 62(4): 389-398.

Whitaker, D. M., Stauffer, D. F., Norman, G. W., Devers, P. K., Edwards, J., Giuliano,

W. M., Harper, C., Igo, W., Sole, J., Spikerand, H., and Tefft, B. 2007. Factors associated with variation in home-range size of Appalachian ruffed grouse (Bonasa umbellus). The

Auk, 124(4): 1407-1424.

Williams, K. E., Hodges, K. E., and Bishop, C. A. 2012. Small reserves around hibernation sites may not adequately protect mobile snakes: the example of Great Basin gophersnakes (Pituophis catenifer deserticola) in British Columbia. Can. J. Zool. 90:

304-312.

Wisler, C., Hofer, U., and Arlettaz, R. 2008. Snakes and monocultures: habitat selection and movements of female grass snakes (Natrix natrix l.) in an agricultural landscape. J.

Herpetol. 42(2): 337-346.

Wright, J. D. 2008. Bullsnake, Pituophis catenifer sayi, nesting biology in Alberta. Can.

Field Nat. 122(2): 138-141.

Zappalorti, R., Burger, J., and Peterson, F. 2015. Home range size and distance traveled from hibernacula in Northern pinesnakes in the New Jersey pine barrens. Herpetologica,

71(1): 26-36.

Chapter 2

Variable habitat selection and space use among bullsnake (Pituophis catenifer sayi)

populations: distance between seasonal habitats drives space use

Introduction

Space use and habitat selection by snakes are based on the spatial distribution of resources in the environment. Primarily, space and habitat use are linked to thermal requirements, with snakes selecting habitats conducive to optimizing thermoregulation

(Burger and Zappalorti 1992; Blouin-Demers and Weatherhead 2001a; Blouin-Demers and Weatherhead 2008; Cross et al. 2015). Retreat sites can be particularly critical, as they provide suitable habitat for thermoregulation, refuge from predators, and increased foraging opportunities (Charland and Gregory 1995; Rodriguez-Robles 2003; Himes et al. 2006; Croak et al. 2013). The ability to locate refuges is especially important for snakes at northern latitudes, where the distribution and abundance of overwintering den sites often limits space use (Burger et al. 1988; Jorgensen et al. 2008; Bauder et al. 2015).

As a result, the proximity of den sites relative to other resources may be a primary determinant of space requirements in northern populations (Martino et al. 2012).

However, few studies have examined snake space use at the northern range limits of a species, where the spatial arrangement of den sites relative to summering habitat may differ.

Resource use by snakes may also be affected in various ways by human modification of the landscape or of specific resources. The removal of native habitat can result in negative impacts on abundance, activity patterns, and behaviors (Burger 2001;

Kjoss and Litvaitis 2001; Beale et al. 2016). However, in many cases, individual responses vary depending on ability to tolerate habitat changes and the quality of resources available (Driscoll 2004; Corey and Doody 2010). For example, individuals may demonstrate increases, decreases, or no difference in the frequency of movements or extent of space use in human modified landscapes compared to natural landscapes (Corey

and Doody 2010; Anguiano and Dieffendorfer 2015; Smith et al. 2015; Ettling et al.

2016). Species may even be positively associated with modified habitats (Carfagno and

Weatherhead 2006; Knoot and Best 2011). Snake response to habitat modification remains to be addressed thoroughly in areas with variation in human land use types, such as agriculture and human recreation, as well as land use intensity.

Snake habitat use and movement patterns vary within and among snake populations. For example, midget-faded rattlesnakes (Crotalus oreganus concolor) exhibit variation in home range size between different populations, as well as between different sexes and reproductive groups within a population (Parker and Anderson 2007).

Prairie rattlesnakes (Crotalus viridis viridis) demonstrate high variability in home range size and movements throughout their geographic range (total distance from 2.76 to 40 km; home range from 18 to 109 ha; summarized by Bauder et al. 2015), with variation in prey availability being the suggested driver for these differences. Habitat selection also varies, with grassland being the most frequently selected habitat, followed by sagebrush steppe and bunchgrass/Douglas fir (summarized by Bauder et al. 2015). Ratsnakes

(Pantherophis spiloides) in Illinois and Ontario select for different habitats, with northern snakes using forest edge habitat more frequently to meet thermoregulatory needs

(Carfagno and Weatherhead 2006). Finally, Gomez et al. (2015) found that two western rattlesnake (Crotalus oreganus) populations, separated by 21 km, demonstrated different movement patterns and habitat selection. Though these previous studies demonstrate differences within and between snake populations, the extent to which spatial ecology may vary between intraspecific populations and the potential drivers of this variation remain to be fully addressed for many species.

When designating protected areas, regulations are typically broad and assume that

populations of the same species have similar requirements. For example, in Canada a

500-m conservation buffer of critical habitat has been designated on federal lands around all den sites of the Threatened eastern yellow-bellied racer (Coluber constrictor flaviventris; Parks Canada Agency 2010). Martino et al. (2012) found that this buffer is not sufficient to protect both summer and winter habitats for racers, at least for the populations in the Frenchman River Valley of Saskatchewan, which make long distance migrations between summer habitat and winter den sites. Williams et al. (2012) also examined the effectiveness of 200 to 300-ha wildlife habitat areas encompassing Great

Basin gophersnake (Pituophis catenifer deserticola) den sites. Though the majority of gophersnakes were protected within the wildlife habitat area, 15% of snakes travelled outside of the area and many snakes did not have home ranges matching the circular shape of the protected area (Williams et al. 2012). The variance in habitat and space use among populations, and how to consider this variance when developing conservation strategies, remains to be examined.

Bullsnakes (Pituophis catenifer sayi) are widespread throughout North America, but remain to be comprehensively examined at their northern range limits, especially in terms of their habitat and space use. Studies from southern range areas show that bullsnake home range size is variable (Fitch 1999; Moriarty and Linck 1997; Rodriguez-

Robles 2003; Kapfer et al. 2008; 2010). Kapfer et al. (2010) found that habitat quality has the largest effect on home range size in this species. Bullsnake habitat selection also varies across their range, with some populations selecting for south facing bluffs (Kapfer et al. 2008), while others select primarily for open grassland habitats (Moriarty and Linck

1997; Rodriguez-Robles 2003). Bullsnakes are currently listed as Data Deficient in

Canada (COSEWIC 2002). Studies concerning space and habitat use have focused on the

Frenchman River Valley of Saskatchewan (Martino et al. 2012; Gardiner et al. 2013). In the Frenchman River Valley, bullsnakes make long distance migrations between summer and winter habitats (up to 4 km), rely heavily on burrow refuge sites, and select for lowland native pasture, slopes, and roads (Martino et al. 2012; Gardiner et al. 2013).

Home ranges in this area were also found to be larger than southerly populations in the

United States (Martino et al. 2012). The characteristics identified from the few previous studies may not be indicative of all bullsnake populations, suggesting the need for a comprehensive, multi-population approach.

Here, I quantify bullsnake space use and habitat selection in three different major river valley systems in Saskatchewan, Canada. My over-arching hypothesis was that habitat availability and landscape configuration affect the spatial distribution of resources in the environment and thus, the space use and habitat selection of snakes. Consequently,

I predicted that bullsnake space and habitat use would vary significantly among populations in the different river valleys, as these areas differ in their available habitat types and landscape configurations, as well as in the intensity of human landscape modification. Comparisons among populations allowed me to assess the extent to which habitat selection and space use vary and to identify common important habitat features between sites. To develop conservation strategies, it is important to understand how snakes use the landscape in various circumstances as well as how human modification to the environment affects habitat use (Harvey and Weatherhead 2006b; Corey and Doody

2010).

Materials and Methods

Study species

The bullsnake (Pituophis catenifer sayi) is one of seven different subspecies of the wide-ranging North American gophersnake (Pituophis catenifer; Ernst and Ernst

2003). It is one of the two subspecies that occur in Canada, the second being the Great

Basin gophersnake (Pituophis catenifer deserticola) found in British Columbia. The bullsnake is the largest species of snake found in Canada, reaching a maximum length of

2.5 meters (Ernst and Ernst 2003; Wright 2016). Bullsnakes are non-venomous constrictors that prey on small mammals, such as mice and ground squirrels, as well as birds, bird eggs, and reptiles. These snakes are diurnal during the majority of the summer, but may become more active during crepuscular periods when conditions are exceedingly hot and dry (Ernst and Ernst 2003). Bullsnakes are widespread throughout the United

States, with their range extending northward into Canada, across southeast Alberta to southwest and southcentral Saskatchewan (Ernst and Ernst 2003). The bullsnake is currently listed as Data Deficient by the Committee on the Status of Endangered Wildlife in Canada (COSEWIC), due to the lack of basic data, including population size, distribution in Canada, and potential threats (COSEWIC 2002).

Study areas

Data on bullsnake space and habitat use were collected from three major valley systems across southern Saskatchewan, Canada: the Frenchman River Valley

(49°10’37”N 107°25’33”W; in 2008 and 2009 by Martino et al. 2012), the Big Muddy

Valley (49°12’55”N 105°12’09”W; in 2015), and the South Saskatchewan River Valley

(°38’16”N 107°59’28”W; in 2016; Figure 1). All data were collected using similar methodology across river valleys. These valleys are located within the mixed grassland eco-region of Saskatchewan and are dominated by native grassland habitats (composing

67% to 91% of the study site), with interspersed shrubs.

Figure 1. Study locations where bullsnakes were tracked using radio-telemetry in southern Saskatchewan: the Frenchman River Valley (FRV, 2008-2009, data collected by Martino et al. 2012), Big Muddy Valley (BMV, 2015), and South Saskatchewan River Valley (SSRV, 2016), indicated by the black pentagons. The North American bullsnake range is highlighted in dark grey.

The Frenchman River Valley site is located within a federally owned PFRA

(Prairie Farm Rehabilitation Administration) community pasture. The PFRA is composed of large tracts of native pasture surrounded by cropland and roads. The native plant community of this area is composed of native grasses (dominated by blue grama

(Bouteloua gracilis), needle and thread (Hesperostipa comata), and western wheat grass

(Agropyron smithii)), and woody vegetation, including silver sagebrush (Artemisia frigida), western snowberry (Symphoricarpos occidentalis), and creeping juniper

(Juniperus horizontalis). The two communal den sites located within the pasture are found on hillsides, with significant hill slumping and large burrow systems (Martino et al.

2012; Gardiner et al. 2013). Bullsnakes have previously been observed to travel from den sites in native upland areas to lowland areas during the summer active season (Martino et al. 2012; Gardiner et al. 2013). As the PFRA program has recently been dissolved in

Saskatchewan, the fate of this area of native habitat is uncertain.

In addition to the native pasture dominating the main valley and adjacent coulees, the Big Muddy Valley houses ranches directly within the valley lowlands. The surrounding uplands have been converted for crop production. The native plant community of this area consists of blue grama, needle and thread, and foxtail barley

(Hordeum jubatum) grasses, with woody vegetation including silver sagebrush, shrubby cinque-foil (Dasiphora fruticosa), creeping juniper, common juniper (Juniperus communis), and wolf willow (Elaeagnus commutate). Bullsnakes have previously been observed to overwinter in the various coulees and slopes near the uplands of the valley.

Den sites are located in the crevice and burrow systems found within various rock formations and hillsides near the valley walls. Bullsnakes have also been observed

extensively throughout the valley. However, no specific habitat associations have been documented, as this population has not previously been studied.

The portion of the South Saskatchewan River within Saskatchewan Landing

Provincial Park (i.e. Lake Diefenbaker) is a popular summer recreational area for campers, anglers, and boaters, bringing in approximately 253,500 visitors per year (D.

Silversides, personal communication, Nov. 8, 2016). Though native prairie dominates the main valley and adjacent coulees, various visitor areas (including campgrounds, golf course, and cottages) are found throughout the base of the valley. Uplands surrounding the park are used for ranching and crop production. Blue grama, needle and thread, and crested wheat grasses (Agropyron cristatum) are prevalent throughout this area, with woody vegetation dominated by silver sagebrush, creeping juniper, common juniper, common snowberry (Symphoricarpos albus), and gooseberry (Ribes grossularia) shrubs.

Den sites throughout the park were found within burrow systems in various topographies, from hillsides to flat, native fields. Bullsnakes are frequently sighted by park visitors throughout Saskatchewan Landing Provincial Park. However, as in the Big Muddy

Valley, no specific habitat associations have been previously documented for this population.

Snake capture and transmitter implantation

Bullsnakes were located during the active season (April to October) via walking and road surveys, and captured by hand in the Frenchman River Valley from 2008-2009, the Big Muddy Valley in 2015, and the South Saskatchewan River Valley in 2016. Upon capture, snakes were measured, weighed, sexed (as in Schaefer 1934), and implanted with passive integrated transponder (pit) tags. Prior to radio-transmitter implantation, snakes were housed for a maximum of 10 days within 16x12x6” Rubbermaid©

containers and provided with water and a heat source. Snakes were transported in these containers to a veterinary clinic (North Albert Veterinary Clinic, Regina for snakes captured in the Big Muddy Valley; Western College of Veterinary Medicine, Saskatoon, for snakes captured in the Frenchman and South Saskatchewan River Valleys) and surgically implanted with Holohil model SI-2, 13-g radio-transmitters (similar to Lentini et al. 2011) by veterinary surgeons. In 2016 implantation protocols were modified.

Whereas the transmitter wire was implanted under the body integument from 2008 to

2015, in 2016 the wire was implanted within the body cavity. Snakes were only implanted if the diameter of the transmitter was less than 50% of the body diameter at the implantation site and the mass of the transmitter was less than 5% of the snake’s body mass. Snakes were allowed a minimum of 24 hours recovery; upon which they were released at their original capture location. All animal handling and surgical procedures were approved by the President’s Committee on Animal Care at the University of Regina

(Animal Care and Use Protocol 13-02).

Radio-telemetry

Following release, efforts were made to locate implanted bullsnakes every 48 hours for the duration of the summer active season. After snakes returned to den sites in the fall, tracking was reduced to once every 1 to 2 weeks. As reptiles tend to use the same location over long periods of time, consistent relocations of snakes every few days are considered sufficient to allow for home range estimation (Row and Blouin-Demers

2006). Upon each location of a snake, Universal Transverse Mercator (UTM) coordinates were recorded on a handheld Global Positioning System (GPS) unit.

Space use and movement patterns

Movement patterns were quantified using ArcGIS 10.3.1 (ESRI 2015). The maximum displacement by individual snakes from their respective den sites was calculated using the Point Distance Tool. The Geospatial Modeling Environment (GME;

Beyer 2015) was used to estimate 100 % minimum convex polygons (MCP) as well as

50% and 95% kernel density estimates (KDE). MCP was calculated for individuals with at least 10 relocations (Himes et al. 2006; Parker and Anderson 2007; Kapfer et al. 2008,

2010; Martino et al. 2012) and KDE for individuals with at least 15 relocations (as in

Gardiner et al. 2013). Relocations are defined as any location to which a snake was tracked. Regression analysis was performed to determine if the number of relocations per snake affected home range size (as in Kapfer et al. 2008). There was no significant positive relationship found between home range size and the number of observations per snake (data not shown).

For the KDE, smoothing factors for each individual snake were estimated using the Plugin algorithm, which calculates the X and Y variances as well as the X/Y covariance among relocation UTM coordinates. I used this method to calculate smoothing factors, as the commonly used Least Squared Cross-Validation (LSCV) algorithm tends to oversmooth and is not recommended for individuals with multiple relocations at the same site (Row and Blouin-Demers 2006). Using the Plugin algorithm to calculate a smoothing factor also produced 95% KDE areas that were most similar to

MCP areas (as recommended in Row and Blouin-Demers 2006), compared to the LSCV.

Home range overlap was calculated for all river valleys using two different methods. The distances between bullsnake MCP centroids were calculated using the

Point Distance Tool and the proportion of bullsnake MCP shared with another snake was calculated using the Intersect Tool in ArcGIS. Distance between centroids and proportion

home range overlap was calculated for bullsnakes occupying the same den sites. To examine the effect of river valley on centroid distance and proportion home range overlap, I performed gamma and beta regression analyses, respectively. I also constructed

Generalized Linear Models (with a gamma distribution) to estimate which variables influence snake home range size (MCP; 95% KDE; 50% KDE) and the maximum displacement by snakes from den sites. Fixed effects were snake sex, snout to vent length, distance to nearest man-made structure (i.e. farmyards, campgrounds, parking lots, cottages), and river valley. Distance to nearest man-made structure was calculated as the distance between the centroid of an individual snake’s MCP and the nearest man- made structure point feature using the Point Distance Tool. After running the global model, Akaike Information Criterion corrected for small samples size (AICc) was used for model selection and competitive models, with ΔAIC <10, were model averaged and the 95% confidence intervals calculated (Burnham et al. 2011).

Third order habitat selection: compositional analysis

Third order habitat selection is defined as selection at the landscape level between habitats distinguishable by larger scale features, such as topography and vegetation type

(Johnson 1980). Habitat type was determined by visually observing the surrounding habitat when a snake was located during tracking. Third order habitat selection was examined using compositional analysis, comparing the number of observations in each habitat type to the proportion of each habitat type available to bullsnakes (Aebischer et al.

1993).

Habitats were considered to be available to a snake if they were contained within an available habitat buffer zone, calculated to be the maximum displacement by snakes from den sites (1.3 km radius buffer for the Big Muddy Valley; 2.4 km radius buffer for

the South Saskatchewan River Valley; as in Gomez et al. 2015). Habitats available to snakes in the Big Muddy Valley included native pasture, crop fields, roads/road areas, hills/slopes/rock formations, trees/tall shrubs, and farmyards. South Saskatchewan River

Valley habitats included native prairie, tame fields, mowed areas, human-developed areas

(such as parking lots, buildings, and campgrounds), crop fields, roads/road areas, beach area, trees/tall shrubs, marshes/streams, and open water. Third order habitat selection by bullsnakes in the Frenchman River Valley was measured by Gardiner et al. (2013), where available habitats included riparian areas, hills/slopes, lowland pasture, mudflats, roads, irrigation areas, native upland, crop fields, prairie dog towns, and open water.

Polygons encompassing available habitats within buffer zones were traced on a high-resolution satellite image (downloaded from flysask2.ca; accessed September 10,

2016) in ArcGIS. The proportion of each habitat within the buffer zone was calculated

(proportion available), as well as the proportion of observations for each individual snake within each habitat type (proportion used). I used the adehabitatHS package in R

(Calenge 2006) to perform compositional analysis to test for non-random habitat selection and rank available habitats based on their selection by bullsnakes.

Fourth order habitat selection

Fourth order habitat selection is defined as the selection of the immediate and local habitat, comprised of physical and ecological features that distinguish it from the surrounding environment (Johnson 1980). A used versus available study design was followed to quantify fourth order habitat selection, where various habitat characteristics are compared between used and available sites (Aebischer et al. 1993; Thomas and

Taylor 2006). Microhabitat characteristics were measured at sites used by snakes and measurements were only taken when a snake was tracked to a new location, defined as

being 20 m or greater from the previous location to which a snake was tracked (similar to

Harvey and Weatherhead 2006a).

Once a snake had vacated an area, a 50 x 50 cm quadrat was placed on the used location and the % grass, forb, shrub, and bare ground cover was estimated (to the nearest

5%) within the quadrat. Maximum vegetation height, distance to nearest burrow, and distance to nearest shrub were also measured. Robel pole measurements of vegetation density were taken in each cardinal direction and averaged (Robel et al. 1970). These habitat variables have been used previously to assess snake habitat selection at a fine scale (Harvey and Weatherhead 2006b; Moore and Gillingham 2006; Martino et al. 2012;

Gardiner et al. 2015). Microhabitat measurements were then repeated at 10 m in each cardinal direction from the used site. This was done to examine whether snakes were selecting habitat at a fine scale (specific to the snake’s site) or at a more local scale

(within a 10 m radius). For analysis, measurements from each of the 4 cardinal directions were averaged to generate one single value for each habitat variable.

These habitat characteristics were also measured at available sites and at 10 m in each cardinal direction from the available site. Available sites were characterized as the halfway point along a straight line between a snake’s previous location and new relocation (>20 m away from previous location; Martino et al. 2012; similar to Gardiner et al. 2015), as this habitat would be ‘available’ to snakes during their travel to a new location. This was done to examine whether snakes were selecting habitat at a local scale

(fourth order habitat selection) within their home range.

Data collected from vegetation surveys were analyzed using resource selection functions (RSFs; Manly et al. 2002). I developed RSF models to compare habitat characteristics between sites used by snakes and those available to them (fourth order

habitat selection, Johnson 1980). I also determined if any of the measured habitat variables affected the probability of a bullsnake selecting for a site at a fine spatial scale.

To model habitat selection, I built Generalized Linear Mixed Models (GLMMs) with a binomial distribution in the package lme4 in R (Bates et al. 2015). Habitat variables were fixed effects and individual snake ID was the random effect. I developed separate models for the Big Muddy and South Saskatchewan River Valleys. Fourth order habitat selection by bullsnakes in the Frenchman River Valley was modeled by Martino et al. (2012).

Prior to running the model, a non-parametric Spearman’s test was used to examine correlations between variables. I eliminated all variables that were correlated greater than R=0.70. As such, % bare ground cover was removed from all models, as it was negatively correlated with % grass cover (R=-0.70 to -0.92). After eliminating correlated variables and running the global GLMM, I used model selection based on

Akaike Information Criterion corrected for small samples size (AICc) to compare all possible combinations of predictor variables. Competitive models, with ΔAIC <10, were model averaged to provide parameter estimates, importance values, and standard errors for all variables (Burnham et al. 2011). The 95% confidence intervals were also calculated for all parameters.

Results

Radio-telemetry

A total of 47 bullsnakes (18 adult females, 17 adult males, 12 undetermined sex) were captured in the Frenchman River Valley from 2008 to 2009. Of these snakes, 14 were radio-tracked over the course of the summer, with the total number of relocations per individual ranging from 10 to 50 (Table 1). In the Big Muddy Valley, 26 bullsnakes were captured from May to July, 2015 (16 adult females, 10 adult males). Of the captured

Table 1. Mean (±SD) telemetry data, maximum displacement, and home range overlap of bullsnakes from three different river valleys in southern Saskatchewan: the Big Muddy Valley (BMV), South Saskatchewan River Valley (SSRV), and the Frenchman River Valley (FRV). N = total number of snakes tracked. Mean maximum Telemetry data Home range overlap displacement River Valley Tracking # From overwintering Distance between Proportion of period relocations den site (m) centroids (m) home range shared (days) FRV (N=14) 62 ± 32.9 25 ± 13.8 1709 ± 959.2 994 ± 502.6 0.14 ± 0.235 BMV (N=7) 102 ± 7 51 ± 3.8 638 ± 380.7 195 ± 123.7 0.49 ± 0.308 SSRV (N=14) 72 ± 26.7 32 ± 11.6 1440 ± 568.0 736 ± 523.6 0.21 ± 0.253

bullsnakes, 7 were radio-tracked over the summer, with total relocations per individual ranging from 43 to 55 (Table 1). In the South Saskatchewan River Valley, a total of 38 bullsnakes were captured from May to August, 2016 (13 adult females, 21 adult males, 4 juvenile males). Of the captured snakes, 14 were implanted with radio-transmitters. The number of relocations per snake ranged from 12 to 48 (Table 1). The maximum time between tracking events was 19 days in the Frenchman River Valley, 7 days in the Big

Muddy Valley, and 6 days in the South Saskatchewan River Valley.

Space use and movement patterns

The top model explaining maximum displacement from den site by bullsnakes included river valley as a fixed effect (Table 2). Snakes in the Frenchman and South

Saskatchewan River Valleys moved similar distances from den sites, with distances in the

Frenchman ranging from 593 to 3946 m and distances in the South Saskatchewan ranging from 661 to 2427 m. Bullsnakes in the Big Muddy Valley, however, tended to move distances 2.3 to 2.7 times shorter, on average, from overwintering den sites than snakes in other areas (Figure 2).

Seven snakes in the Big Muddy Valley, 14 snakes in the Frenchman River Valley, and 14 snakes in the South Saskatchewan River Valley were relocated often enough to estimate MCP home range area, while 7 Big Muddy, 10 Frenchman, and 13 South

Saskatchewan snakes had enough relocations to estimate the 50% and 95% KDEs. The top model explaining bullsnake home range size, regardless of home range estimator, included river valley (Table 3, 4, and 5). On average, bullsnakes in the Frenchman and

South Saskatchewan River Valleys had MCP home ranges that were 3 to 3.7 times larger,

95% KDEs that were 4 to 4.5 times larger, and 50% KDEs that were 4.3 to 4.4 times larger than those in the Big Muddy Valley (Figure 3). Home range overlap was greater on

Table 2. Top generalized linear model, null model, and all models with ΔAIC <3 evaluating the best predictor of the maximum displacement from overwintering den sites by Saskatchewan bullsnakes. Fixed effects included river valley, distance to nearest man- made structure (m), snout to vent length (SVL, cm), and snake sex. The number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights Intercept only 1 501.50 9.57 0.00 AIC model valley + dist. man-made 2 491.92 0.00 0.37 selection valley 1 492.95 1.03 0.22 valley + dist. man-made + SVL 3 494.07 2.15 0.13 Parameter Estimate SE Lower 95% CI Upper 95% CI Importance values (Intercept) 6.34 0.42 5.62 7.05 NA valley frv 1.21 0.34 0.64 1.78 0.99 Model 0.99 averaging valley ssrv 0.88 0.27 0.43 1.34 dist. m 0.00 0.00 0.00 0.00 0.61 svl 0.00 0.00 0.00 0.01 0.26 sex M 0.00 0.09 -0.16 0.16 0.19

Figure 2. Minimum convex polygons (MCPs) for bullsnakes in the Frenchman River (FRV), South Saskatchewan River (SSRV), and Big Muddy (BMV) Valleys. MCPs are shown for 9 Frenchman River Valley, 8 South Saskatchewan River Valley, and 4 Big Muddy Valley bullsnakes. Den sites are indicated by stars, roads by thick black lines, urban areas by dark grey polygons, crop fields by crosshatched polygons, and lakes by light grey polygons. Scale differs among panels.

Table 3. Top generalized linear model, null model, and all models with ΔAIC <3 evaluating the best predictor of Saskatchewan bullsnake 50% KDE core area size. Fixed effects included river valley, distance to nearest man-made structure (m), snout to vent length (SVL, cm), and snake sex. The number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights Intercept only 1 176.53 6.79 0.02 AIC model valley + SVL 2 selection 169.74 0.00 0.50 valley 1 172.19 2.45 0.15 Parameter Estimate SE Lower 95% CI Upper 95% CI Importance values (Intercept) 3.13 1.38 0.82 5.44 NA valley frv 1.87 0.63 0.82 2.93 0.96 Model 0.96 averaging valley ssrv 1.61 0.54 0.69 2.53 svl -0.02 0.01 -0.04 0.00 0.79 dist. m 0.00 0.00 0.00 0.00 0.24 sex M -0.06 0.24 -0.46 0.34 0.20

Table 4. Top generalized linear model, null model, and all models with ΔAIC <3 evaluating the best predictor of Saskatchewan bullsnake 95% KDE home range area size. Fixed effects included river valley, distance to nearest man-made structure (m), snout to vent length (SVL, cm), and snake sex. The number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights Intercept only 1 266.50 7.24 0.01 AIC model valley + SVL 2 selection 259.42 0.00 0.39 valley 1 261.43 2.16 0.13 Parameter Estimate SE Lower 95% CI Upper 95% CI Importance values (Intercept) 4.00 1.18 2.03 5.97 NA valley frv 1.71 0.53 0.81 2.60 0.97 Model 0.97 averaging valley ssrv 1.32 0.44 0.57 2.07 svl -0.01 0.01 -0.03 0.01 0.63 dist. m 0.00 0.00 0.00 0.00 0.28 sex M -0.09 0.24 -0.49 0.31 0.25

Table 5. Top generalized linear model, null model, and all models with ΔAIC <3 evaluating the best predictor of Saskatchewan bullsnake MCP home range size. Fixed effects included river valley, distance to nearest man-made structure (m), snout to vent length (SVL, cm), and snake sex. The number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights are presented. Model averaging was performed, with presented values including the different model parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights Intercept only 1 295.94 4.67 0.04 valley 1 291.27 0.00 0.45 AIC model valley + sex 2 selection 293.79 2.52 0.13 valley + dist. man-made 2 293.98 2.71 0.12 valley + SVL 2 294.11 2.84 0.11 Parameter Estimate SE Lower 95% CI Upper 95% CI Importance values (Intercept) 2.96 0.58 1.97 3.95 NA valley frv 1.20 0.57 0.25 2.15 0.90 Model 0.90 averaging valley ssrv 1.06 0.50 0.22 1.90 sex M -0.04 0.16 -0.30 0.23 0.21 dist. m 0.00 0.00 0.00 0.00 0.20 svl 0.00 0.00 -0.01 0.01 0.19

Figure 3. Box and Whisker Plot of home ranges for bullsnakes in the Frenchman River Valley (FRV; N=14), Big Muddy Valley (BMV; N=7), and South Saskatchewan River Valley (SSRV; N=14). The median is displayed as the thick black bar; boxes represent the first and third quartiles; whiskers show the highest and lowest values within 1.5 times the interquartile range. Dots indicate outliers. The 100% Minimum Convex Polygon (MCP) is shown in dark grey, the 95% Kernel Density Estimate in light grey, and the 50% Kernel Density Estimate in white.

average in the Big Muddy Valley, compared to the Frenchman and South Saskatchewan

River Valleys (Table 1; Figure 2). This was supported by the regression analyses, which demonstrated that distance between centroids increased and the proportion of home range overlap decreased in the Frenchman and South Saskatchewan River Valleys compared to the Big Muddy Valley (Table 6).

Third order habitat selection

Bullsnakes in the Big Muddy Valley and South Saskatchewan River Valley demonstrated third order habitat selection, with variation in selected habitats among river valleys. Bullsnakes in the Big Muddy Valley exhibited non-random habitat use (λ=0.002, p=0.028), with the most frequently used habitat types being farmyards and native pasture.

On average, farmyards were selected 11 times more than expected, while native pasture was used based on availability (Figure 4). Roads and hills/slopes were also used based on availability, and crop fields and treed areas were not used at all (Figure 4).

Bullsnakes in the South Saskatchewan River Valley also exhibited non-random habitat use (λ=0.014, p=0.01), with the most frequently used habitats being beach area, native prairie, tame fields, human-developed areas (including buildings, parking lots, and campgrounds), mowed areas, and roads. Bullsnakes used beaches 91.6 times, tame fields

8.7 times, buildings 2.1 times, mowed areas 2.9 times, and roads 2.3 times more than expected based on availability (Figure 5). Native prairie and marshes were used based on availability (Figure 5). Treed areas and crop fields were used 8.7 times and 2 times less than expected (Figure 5). Snakes were not observed in open water; however, they did make use of this habitat to travel from one side of Lake Diefenbaker to the other.

Table 6. Regression analyses evaluating the differences in home range overlap between the Frenchman River (FRV), Big Muddy (BMV), and South Saskatchewan River (SSRV) Valleys. Home range overlap was measured in two ways: as the distance between minimum convex polygon centroids and the proportion of home range overlap between snakes. A gamma regression was used to examine the effect of river valley on centroid distance. A beta regression was used examine the effect of river valley on proportion of home range overlap. The fixed effect for both models was river valley. Presented here are the parameter estimates, standard errors (SE), t-values/z-values, and p-values. Centroid distance ~ river valley Parameter Estimate SE t-value p-value valley FRV 799.08 96.25 8.30 <0.001 valley SSRV 540.89 98.19 5.51 <0.001 Proportion of home range overlap ~ river valley Parameter Estimate SE z-value p-value valley FRV -1.66 0.30 -5.50 <0.001 valley SSRV -1.39 0.32 -4.31 <0.001

100

90 Used 80 Available 70

40 Percent (%) (%) Percent 30

0 Crop Trees Roads Hills Pasture Farm yard

Figure 4. Percent of different habitats used and available within a 1.3 km buffer surrounding den sites in the Big Muddy Valley, Saskatchewan.

100

90 Used 80 Available 70

40 Percent (%) (%) Percent 30

0 Marsh Trees Crop Road Mowed Urban Tame Native Beach

Figure 5. Percent of different habitat types used and available within a 2.4 km buffer surrounding den sites in the South Saskatchewan River Valley within Saskatchewan Landing Provincial Park, Saskatchewan.

Fourth order habitat selection

The top model explaining differences between used and available sites in the Big

Muddy Valley and South Saskatchewan River Valley included % grass cover, vegetation density, and distance to the nearest burrow (Table 7 and 8). The probability of locating a snake increased with decreasing grass cover, increasing vegetation density, and decreasing distance to the nearest burrow. The model-averaged 95% confidence intervals for distance to nearest burrow, % grass cover, and vegetation density did not pass through zero for both river valleys and the importance values for these three variables were greater than 0.8 (Table 7 and 8). Bullsnakes were most likely to be found within 1 m of a burrow or other refuge site (Figure 6).

At the local scale (10 m in each cardinal direction at used and available sites), the distance to nearest burrow was an important predictor of fourth order habitat selection by bullsnakes in the Big Muddy and South Saskatchewan River Valleys, and was the only variable included in the top models (Table 9 and 10). All other variables had no major effect on bullsnake local scale habitat selection. The model-averaged 95% confidence intervals for distance to nearest burrow did not pass through zero (Table 9 and 10) and the importance value for this variable was 1, indicating that proximity to a burrow was an important predictor of bullsnake habitat selection at the local scale.

Discussion

Bullsnake space use and movement patterns vary across their geographic range.

Home range areas estimated here are comparable to or larger than those observed for bullsnake populations in southern range areas. Saskatchewan bullsnake home ranges varied from 3 to 175 ha (MCP), compared to the previous maximum home range for this

Table 7. Top generalized linear mixed model, null model, and all models with ΔAIC <2 used for evaluating bullsnake habitat selection in the Big Muddy Valley. Selection was examined using radio-telemetry locations compared to available locations. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), maximum vegetation height (cm), and Robel pole vegetation density. Random effect was individual snake ID. Presented here are the number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights Intercept + |ind| 2 373.88 108.72 0.00 robel + % grass + burrow + |ind| 4 265.17 0.00 0.24 AIC dist. shrub + robel + % grass + burrow + |ind| 5 266.58 1.42 0.12 model selection max veg + robel + % grass + burrow + |ind| 5 266.89 1.72 0.10 % forb + robel + % grass + burrow + |ind| 5 267.01 1.84 0.10 % shrub + robel + % grass + burrow + |ind| 5 267.25 2.09 0.08 Lower Importance Parameter Estimate SE 95% CI Upper 95% CI values (Intercept) 1.60 0.35 1.02 2.18 NA burrow -0.51 0.09 -0.66 -0.37 0.99 Model grass -0.02 0.01 -0.03 -0.01 0.99 averaging robel 3.83 1.89 0.71 6.95 0.98 dist. shrub 0.01 0.01 -0.02 0.03 0.33 max. veg 0.00 0.01 -0.01 0.02 0.30 forb 0.00 0.01 -0.02 0.02 0.27 shrub 0.00 0.01 -0.02 0.02 0.26

Table 8. Top generalized linear mixed model, null model, and all models with Delta AIC <2 used for evaluating bullsnake habitat selection in the South Saskatchewan River Valley. Selection was examined using radio-telemetry locations compared to available locations. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), maximum vegetation height (cm), and Robel pole vegetation density measurements. Random effect was individual snake ID. Presented here are the number of model parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (ΔAIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights intercept + |ind| 2 539.07 123.86 0.00 burrow + % forb + % grass + max veg + robel + |ind| 6 415.24 0.00 0.11 burrow + % grass + max veg + robel + |ind| 5 415.31 0.07 0.11 burrow + % forb + %grass + robel + |ind| 5 415.44 0.20 0.10 AIC burrow + % grass + max veg + robel + % shrub + |ind| 6 415.78 0.54 0.09 model selection burrow + % forb + % grass + max veg + robel + % shrub + |ind| 7 416.23 0.99 0.07 burrow + % forb + % grass + robel + % shrub + |ind| 6 416.53 1.30 0.06 burrow + % grass + robel + |ind| 4 416.81 1.57 0.05 burrow + dist. shrub + % grass + max veg + robel + |ind| 6 417.16 1.92 0.04 burrow + dist. shrub + % forb + % grass + max veg + robel +|ind| 7 417.18 1.95 0.04 Importance Parameter Estimate SE Lower 95% CI Upper 95% CI values (Intercept) 0.98 0.35 0.41 1.55 NA burrow -0.16 0.02 -0.19 -0.13 0.99 Model grass -0.01 0.01 -0.02 0.00 0.89 averaging robel 0.81 0.29 0.33 1.29 0.98 max. veg 0.01 0.01 0.00 0.02 0.58 forb 0.02 0.02 -0.02 0.05 0.56 shrub 0.00 0.01 -0.01 0.01 0.38 dist. shrub 0.00 0.01 -0.01 0.01 0.27

0,9

0,8 BMV SSRV 0,7

0,6

0,5

0,4

0,3 Frequency use of Frequency 0,2

0,1

0 0 to 1 2 to 5 > 6 Distance to nearest burrow (m)

Figure 6. Relative frequency of habitat use by bullsnakes in the Big Muddy Valley (BMV) and South Saskatchewan River Valley (SSRV) at locations 0 to 1 m, 2 to 5 m, and greater than 6 m from a burrow.

Table 9. Top generalized linear mixed model, null model, and all models with Delta AIC <2 used for evaluating bullsnake habitat selection at the local scale. Selection was examined using measurements taken 10 m in each cardinal direction at used and available sites in the Big Muddy Valley. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), Robel pole vegetation density measurements, and maximum vegetation height (cm). Random effect was individual snake ID. Presented here are the number of parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (Delta AIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights intercept + |ind| 2 371.20 15.17 0.00 burrow + robel + % shrub + |ind| 4 356.06 0.00 0.06 burrow + robel + |ind| 3 356.08 0.02 0.06 burrow + |ind| 2 356.10 0.04 0.06 burrow + max veg + robel + % shrub + |ind| 5 357.02 0.97 0.04 burrow + % forb + robel + |ind| 4 357.26 1.21 0.03 AIC burrow + max veg + |ind| 3 357.29 1.23 0.03 model burrow + % forb + |ind| 3 357.29 1.24 0.03 selection burrow + dist. shrub + |ind| 4 357.45 1.39 0.03 burrow + max veg + % shrub + |ind| 4 357.47 1.42 0.03 burrow + % grass + |ind| 3 357.48 1.42 0.03 burrow + % forb + robel + % shrub + |ind| 5 357.60 1.54 0.03 burrow + dist. shrub + robel + |ind| 4 357.68 1.63 0.03 burrow + % grass + robel + |ind| 4 357.76 1.70 0.03 burrow + % shrub + |ind| 3 357.82 1.76 0.02 Importance Parameter Estimate SE Lower 95% CI Upper 95% CI values (Intercept) 0.48 0.35 -0.10 1.06 NA burrow -0.13 0.04 -0.19 -0.06 1.00 Model robel 0.87 1.21 -0.12 2.85 0.52 averaging shrub -0.02 0.03 -0.06 0.03 0.42 max. veg 0.01 0.01 -0.01 0.03 0.36 forb 0.01 0.02 -0.02 0.03 0.34 dist. shrub 0.00 0.01 -0.01 0.02 0.31 grass 0.00 0.00 -0.01 0.01 0.30

Table 10. Top generalized linear mixed model, null model, and all models with Delta AIC <2 used for evaluating bullsnake habitat selection at the local scale. Selection was examined using habitat measurements taken 10 m in each cardinal direction at used and available sites in the South Saskatchewan River Valley. Fixed effects included % grass cover, % shrub cover, % forb cover, distance to nearest burrow (m), distance to nearest shrub (m), Robel pole vegetation density measurements, and maximum vegetation height (cm). Random effect was individual snake ID. Presented here are the number of parameters (K), Akaike’s Information Criterion value corrected for small sample size (AICc), difference in AICc from the top model (Delta AIC), and the Akaike weights. Model averaging was performed, with presented values including the different model, parameter estimates, standard errors (SE), the upper and lower 95% confidence intervals (CI), and the importance values (calculated using the Akaike weights). Model K AICc ΔAIC Weights intercept + |ind| 2 547.17 64.15 0.00 burrow + % grass + robel + % shrubs + |ind| 5 483.06 0.00 0.08 burrow + robel + % shrub + |ind| 4 483.32 0.27 0.07 AIC burrow + robel + |ind| 3 483.74 0.68 0.06 model burrow + % shrub + |ind| 3 483.97 0.91 0.05 selection burrow + |ind| 2 483.99 0.94 0.05 burrow + dist. shrub + % grass + robel + % shrub + |ind| 6 484.87 1.81 0.03 burrow + max veg + robel + % shrub + |ind| 5 484.91 1.86 0.03 burrow + % grass + robel + |ind| 4 485.01 1.95 0.03 Parameter Estimate SE Lower 95% CI Upper 95% CI Importance values (Intercept) 1.64 0.34 1.08 2.20 NA burrow -0.13 0.02 -0.16 -0.10 1.00 Model robel 0.30 0.32 -0.23 0.83 0.62 averaging shrub -0.01 0.01 -0.02 0.01 0.58 grass 0.00 0.00 -0.01 0.01 0.40 dist. shrub 0.00 0.01 -0.01 0.01 0.27 max. veg 0.00 0.00 -0.01 0.01 0.29 forb 0.00 0.01 -0.02 0.02 0.28

species of 34 ha (Fitch 1999; Moriarty and Linck 1997; Rodriguez-Robles 2003; Kapfer et al. 2008). Home range areas appeared to vary among bullsnakes due to similar variation in the maximum distance bullsnakes travelled away from den sites during the summer active season. Maximum distance bullsnakes travelled away from den sites ranged from 352 m to 3.9 km across the three study sites. Variation in movement and space use has also been documented for populations of other snake species. For example,

Bauder et al. (2015) summarized prairie rattlesnake maximum displacement from den sites throughout their geographic distribution, which ranged from 1.5 to 17 km. Gomez et al. (2015) also observed midget-faded rattlesnake populations, separated by only 21 km, where individuals at one site moved significantly farther from den sites than those at the other site. Similarities among separate populations have also been shown, with broad- headed snake (Hoplocephalus bungaroides) populations, separated by approximately 250 km, demonstrating similar home range areas and movement distances (Croak et al. 2013).

It appears that snake spatial requirements may vary greatly across their range, even for populations in relatively close proximity. As such, my results suggest that a single strategy for conservation and management may not suffice, even for a specific area within the geographic range of a species (e.g. southern Saskatchewan).

The top predictor of home range area and distances travelled by bullsnakes was river valley, suggesting that these features of bullsnake spatial ecology differ by river valley. Bullsnake range area and distance travelled in the Frenchman and South

Saskatchewan River Valleys were more similar, using up to 4 times more space on average, and travelling up to 2 times farther from den sites, than those in the Big Muddy

Valley. When summer resource requirements are not met in areas immediately surrounding overwintering sites, snakes are required to travel away from den sites and

thus, use more space. This was observed for bullsnakes in the Frenchman and South

Saskatchewan River Valleys, with summering and overwintering activity centers (in terms of the 95% KDE) being separate, resulting in seasonal migrations (as described in

Gardiner et al. 2013). Alternatively, if summer and winter resource requirements are met in one small area, snakes may occupy relatively smaller home range areas, as was observed for bullsnakes in the Big Muddy Valley. Small space requirements have previously been documented for snakes when prey availability is high in an area (Brown et al. 2005; Ettling et al. 2016), with snakes migrating seasonally to locate prey when this is not the case (Duvall et al. 1990). However, Kapfer et al. (2010) suggested that bullsnake space use is not primarily driven by prey availability, but instead is determined by thermoregulatory and refuge needs.

Regardless of driving factor, higher resource abundance and availability in the

Big Muddy Valley corresponds with the home range overlap data presented here (Figure

3). Bullsnakes in the Big Muddy Valley had higher home range overlap, in addition to smaller home range sizes, suggesting that resources are more abundant and available in close proximity to den sites. This was also supported by body condition data collected across study sites. Big Muddy Valley snakes were in better body condition compared to the other valleys (A. Gallon, unpublished data). Overall, the placement of den sites in relation to summer resources of adequate availability and abundance appears to be an important determinant of bullsnake space use. Although this study demonstrated differences in bullsnake spatial ecology among river valleys, the precise nature of the summer resources driving the movement away from overwintering den sites remain to be determined.

Bullsnake summer habitat use was similar across the South Saskatchewan River and Big Muddy valleys, but ultimately appears to be site-specific. Native habitats were those most frequently used by snakes in the Big Muddy and South Saskatchewan River

Valleys, in addition to a range of human-modified habitats (Figure 3 and 4). Likewise, bullsnakes in the Frenchman River Valley also selected primarily for native habitats, in addition to roads (Gardiner et al. 2013). Native habitats were used as expected based on availability across all valleys, but human-modified habitats were used at different frequencies across populations (2 times more than expected in the Big Muddy and South

Saskatchewan River Valleys, and 3 times less in the Frenchman River Valley). Though previous studies in the southern range areas of the bullsnake have found this species to select primarily for native grassland habitats (Moriarty and Linck 1997; Rodriguez-

Robles 2003), Kapfer et al. (2008) found snakes to select for open bluff areas. Other snake species have demonstrated a similar diversity in habitat selection among populations (Gomez et al. 2015; Bauder et al. 2015). These results suggest that bullsnake habitat requirements differ across populations. This has previously been documented for other snake species, where ratsnakes (Elaphe obsoleta) in Ontario selected for edge habitats more frequently than in southern populations to optimize thermoregulation, as they occupy a more thermally harsh environment (Blouin-Demers and Weatherhead

2001b; Carfagno and Weatherhead 2006; Sperry and Weatherhead 2009). These results also suggest that bullsnakes are highly flexible in terms of their broad scale habitat selection and how they meet resource demands across their large and diverse geographic range. As such, the differential selection of habitats by bullsnakes demonstrated here and in southern areas may be the result of variation in resource requirements as well as resource acquisition in terms of third order habitat selection.

Differential habitat selection among populations may also be the result of differences in habitat availability and habitat diversity across areas. For example, available habitats differed among Saskatchewan river valleys, with bullsnakes in the

Frenchman and South Saskatchewan River Valleys having 10 different habitats available

(differing in dominant substrate/vegetation type, vegetation structure, and degree of human modification), while snakes in the Big Muddy Valley had only six. This suggests that the available habitat composition in one area may not be comparable to that available in another. Similarly, specific habitat types available in one area may not be available in others. As such, snake habitat selection will vary among populations depending, at least partially, on habitat availability and relative abundances. It appears that habitat selection in snakes differs among populations, being dependent on the composition of available habitats and the versatility of the population in terms of how they meet their resource requirements.

Retreat sites are an important habitat feature for bullsnakes. Snakes in the Big

Muddy and South Saskatchewan River Valleys selected sites in close proximity to retreat sites. This was regardless of other habitat features or whether the refuge itself was natural

(mammal burrow) or man-made (under walkways, cement pads, stacked rocks).

Bullsnakes in the Frenchman River Valley also demonstrated a dependence on retreat sites, as did eastern yellow-bellied racers (Coluber constrictor flaviventris) and prairie rattlesnakes (Crotalus viridis viridis; Martino et al. 2012; Gardiner et al. 2015). Suitable retreat sites are an important habitat feature for snakes, as they provide refuge from extreme temperatures and benefit thermoregulation (Huey et al. 1989; Webb and Shine

1998; Himes et al. 2006; Blouin-Demers and Weatherhead 2008; Croak et al. 2013). In this study, bullsnakes remained near retreat sites (within 1m), even when snakes were not

located directly within the retreat site itself. This would be beneficial for thermoregulation, as it would allow snakes to move in and out of refuges, depending on their physiological and thermoregulatory needs (Blouin-Demers and Weatherhead

2001a). Remaining in close proximity to retreat sites, particularly burrows, would also benefit bullsnakes in terms of predator avoidance and foraging (Moriarty and Linck 1997;

Rodriguez-Robles 2002; Heard et al. 2004; Himes et al. 2006). Of those retreat sites used by bullsnakes, many were excavated by mammals. Large burrow networks may also be used as overwintering sites (this study; Moriarty and Linck 1997). As a result, snake reliance on mammal-created refuge sites suggests that the presence of species such as the

Richardson’s ground squirrel (Urocitellus richardsonii), Thirteen-lined ground squirrel

(Ictidomys tridecemlineatus), Nuttall’s cottontail (Sylvilagus nuttallii), and American badger (Taxidea taxus) should be considered when developing conservation and management strategies for bullsnakes.

Conclusions

Bullsnake habitat and space requirements vary among populations as well as across their geographic range. Much of the time, data pertaining to species habitat and space use are based on a single population or study site (Croak et al. 2013). Here, I highlight the importance of understanding the spatial ecology and habitat selection of different populations of a single species where resource requirements and critical habitat features differ greatly among populations. These findings are also relevant to other snake species, which demonstrate a similar variability in habitat and space use (Jorgensen et al.

2008; Bauder et al. 2015; Gomez et al. 2015). As such, conservation and management strategies developed for one population of a species may not be applicable to others. A possible approach for implementing a more inclusive strategy would be the creation of

protected buffers around den sites based on the largest measured space use requirements for a species. Such an approach should encompass both wide and narrow-ranging individuals and populations.

Bullsnake space use during the summer active season differed among river valleys. Third order habitat selection also differed among populations, with bullsnakes appearing to be quite generalist in their habitat requirements and making use of a wide range of habitat types in this as well as previous studies (Moriarty and Linck 1997;

Rodriguez-Robles 2003; Kapfer et al. 2008; Gardiner et al. 2013). The spatial association among seasonal habitats appears to be an important determinant of bullsnake space use.

However, the specific summer resource requirements driving the seasonal migrations of bullsnakes remain to be addressed. Measuring how summer resource abundance and availability varies among populations may be useful in uncovering the drivers of bullsnake summer space and habitat use. At the local scale, retreat sites appear to be a universally important habitat feature. Refuge type (i.e. natural or man-made) did not appear to be of importance. In the future, I recommend examining the thermal properties of selected refuge sites to better understand bullsnake local scale habitat selection and requirements. I also recommend considering the status of co-occurring fossorial mammal populations when developing management strategies for bullsnakes. Fossorial mammals are clearly important in providing food and refuges. Overall, habitat features at the fine spatial scale appear to be an important determinant of bullsnake habitat use, compared to habitat features at the landscape scale.

References

Aebischer, N. J., Robertson, P. A., and Kenward, R. E. 1993. Compositional analysis of habitat use from animal radio-tracking data. Ecology, 74(5): 1313-1325.

Anguiano, M. O. and Diffendorfer, J. E. 2015. Effects of fragmentation on the spatial ecology of the California kingsnake (Lampropeltis californiae). J. Herpetol. 49(3): 420-

427.

Bates, D., Maechler, M., Bolker, B., and Walker, S. 2015. Fitting linear mixed-effects models using lme4. J. Stat. Softw. 67(1): 1-48.

Bauder, J. M., Akenson, H., and Peterson, C. R. 2015. Movement patterns of prairie rattlesnakes (Crotalus v. viridis) across a mountainous landscape in a designated wilderness area. J. Herpetol. 49(3): 377-387.

Beale, M., Poulin, S., Ivanyi, C., and Blouin-Demers, G. 2016. Anthropogenic disturbance affects movement and increases concealment in western diamondback rattlesnakes (Crotalus atrox). J. Herpetol. 50(2): 216-221.

Beyer, H. L. 2015. Geospatial Modelling Environment (Version 0.7.4.0) [software]. http://www.spatialecology.com/gme

Blouin-Demers, G. and Weatherhead, P. J. 2001a. An experimental test of the link between foraging, habitat selection, and thermoregulation in black rat snakes Elaphe

obsoleta obsoleta. J. Anim. Ecol. 70: 1006-1013.

Blouin-Demers, G. and Weatherhead, P. J. 2001b. Habitat use by black rat snake (Elaphe obsoleta obsoleta) in fragmented forests. Ecology, 82(10): 2882-2896.

Brown, G. P., Shine, R., and Madsen, T. 2005. Spatial ecology of slatey-grey snakes

(Stegonotus cucullatus, Colubridae) on a tropical Australian floodplain. J. Trop. Ecol.

21(6): 605-612.

Burger, J. 2001. The behavioral response of basking northern water (Nerodia sipedon) and eastern garter (Thamnophis sirtalis) snakes to pedestrians in a New Jersey park.

Urban Ecosyst. 5: 199-129.

Burger, J. and Zappalorti, R. T. 1992. Philopatry and nesting phenology of pine snakes

Pituophis melanoleucus in the New Jersey pine barrens. Behav. Ecol. Sociobiol. 30(5):

331-336.

Burger, J., Zappalorti, R. T., Gochfeld, M., Boarman, W. I., Caffrey, M., Doig, V.,

Garber, S. D., Lauro, B., Mikovsky, M., Safina, C., and Saliva, J. 1988. Hibernacula and summer den sites of pine snakes (Pituophis melanoleucus) in the New Jersey pine

barrens. J. Herpetol. 22(4): 425-433.

Calenge, C. 2006. The package adehabitat for the R software: a tool for the analysis of space and habitat use by animals. Ecol. Modell. 197: 516-519.

Carfagno, G. L. F. and Weatherhead, P. J. 2006. Intraspecific and interspecific variation in use of forest-edge habitat by snakes. Can. J. Zool. 84: 1440-1452.

Charland, M. B. and Gregory, P. T. 1995. Movements and habitat use in gravid and nongravid female garter snakes (Colubridae: Thamnophis). J. Zool. 236: 543-

561.

Committee on the Status of Endangered Wildlife in Canada (COSEWIC). 2002.

COSEWIC assessment and status report on the Gophersnake Pituophis catenifer in

Canada. Ottawa. pp. 1- 33.

Corey, B. and Doody, J. S. 2010. Anthropogenic influences on the spatial ecology of a semi-arid python. J. Zool. 281: 293-302.

Croak, B. M., Crowther, M. S., Webb, J. K., and Shine, R. 2013. Movements and habitat use of an endangered snake, Hoplocephalus bungaroides (Elapidae): implications for conservation. PLoS ONE. 8(4): e61711.

Cross, M. D., Root, K. V., Mehne, C. J., Mcgowan-Stinski, J., and Pearsall, D. 2015.

Multi-scale response of eastern massasauga rattlesnakes (Sistrurus catenatus) to prescribed fire. Am. Midl. Nat. 173: 346-362.

Driscoll, D. A. 2004. Extinction and outbreaks accompany fragmentation of a reptile community. Ecol. Appl. 14(1): 220-240.

Duvall, D., Goode, M. J., Hayes, W. K., Leonhardt, J. K., and Brown, D. G. 1990. Prairie rattlesnake vernal migration: field experimental analyses and survival value. Natl. Geog.

Res. 6(4): 457-469.

Ernst, C. H. and Ernst, E. M. 2003. Snakes of the United States and Canada. Smithsonian

Books, Washington, D.C., USA. pp. 272-280.

ESRI (Environmental Systems Research Institute). 2015. ArcGIS Desktop. Redlands,

CA.

Ettling, J. A., Aghasyan, L. A., Aghaysan, A. L., and Parker, P. G. 2016. Spatial ecology of Armenian vipers, Montivipera raddei, in two different landscapes: human-modified and recovered-natural. Russ. J. Herpetol. 23(2): 93-102.

Gardiner, L. E., Somers, C. M., Martino, J. A., Parker, D. L., and Poulin, R. G. 2013.

Balancing the dumbbell: summer habitats need protection in addition to winter dens for northern snake communities. J. Wildl. Manage. 77(5): 975-982.

Gardiner, L. E., Somers, C. M., Parker, D. L., and Poulin, R. G. 2015. Microhabitat selection by prairie rattlesnakes (Crotalus viridis) at the northern extreme of their geographic range. J. Herpetol. 49(1): 131-137.

Gomez, L., Larsen, K. W., and Gregory, P. T. 2015. Contrasting patterns of migration and habitat use in neighboring rattlesnake populations. J. Herpetol. 49(3): 371-376.

Hammermeister, A. M., Gauthier, D., and McGovern, K. 2001. Saskatchewan’s native prairie: statistics of a vanishing ecosystem and dwindling resource. Native Plant Society of Saskatchewan Inc., Saskatoon, SK.

Harvey, D. S. and Weatherhead, P. J. 2006a. Hibernation site selection by eastern massasauga rattlesnakes (Sistrurus catenatus catenatus) near their northern range limit. J.

Herpetol. 40(1): 66-73.

Harvey, D. S. and Weatherhead, P. J. 2006b. A test of the hierarchical model of habitat selection using eastern massasauga rattlesnakes (Sistrurus c. catenatus). Biol. Conserv.

130: 206-216.

Heard, G. W., Black, D., and Robertson, P. 2004. Habitat use by the inland carpet python

(Morelia spilota metcalfei: Pythonidae): seasonal relationships with habitat structure and prey distribution in a rural landscape. Aust. Ecol. 29: 446-460.

Hecnar, S. J. and M’Closkey, R. T. 1998. Effects of human disturbance on five-lined

skink, Eumeces fasciatus, abundance and distribution. Biol. Conserv. 85: 213-222.

Himes, J. G., Hardy, L. M., Rudolph, D. C., and Burgdorf, S. J. 2006. Movement patterns and habitat selection by native and repatriated Louisiana pine snakes (Pituophis ruthveni): implications for conservation. Herpetol. Nat. Hist. 9(2): 103-116.

Huey, R. B., Peterson, C. R., Arnold, S. J., and Porter, W. P. 1989. Hot rocks and not-so- hot rocks: retreat-site selection by garter snakes and its thermal consequences. Ecology,

70(4): 931-944.

Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology, 61(1): 65-71.

Jorgensen, D., Gates, C. C., and Whiteside, D. P. 2008. Movements, migrations, and mechanisms: a review of radiotelemetry studies of prairie (Crotalus v. viridis) and western (Crotalus oreganus) rattlesnakes. In: Hayes, W. K., Beaman, K. R., Cardwell,

M. D., and Bush, S. P. The biology of rattlesnakes. Loma Linda University Press, Loma

Linda, California. pp. 303-316.

Kapfer, J. M., Coggins, J. R., and Hay, R. 2008. Spatial ecology and habitat selection of bullsnakes (Pituophis catenifer sayi) at the northern periphery of their geographic range.

Copeia, 2008(4): 815-826.

Kapfer, J. M., Pekar, C. W., Reineke, D. M., Coggins, J. R., and Hay. R. 2010. Modeling

the relationship between habitat preferences and home-range size: a case study on a large mobile Colubrid snake from North America. J. Zool. 282: 13-20.

Kjoss, V. A. and Litvaitis, J. A. 2001. Community structure of snakes in a human- dominated landscape. Biol. Conserv. 98: 285-292.

Knoot, T. G. and Best, L. B. 2011. A multiscale approach to understanding snake use of conservation buffer strips in an agricultural landscape. Herpetol. Conserv. Biol. 6(2):

191-201.

Lentini, A. M., Crawshaw, G. J., Licht, L. E., and McLelland, D. J. 2011. Pathologic and hematologic responses to surgically implanted transmitters in eastern massasauga rattlesnakes (Sistrurus catenatus catenatus). J. Wildl. Dis. 47(1): 107-125.

Manly, B. F. J., McDonald, L. L., Thomas, D. L., McDonald, T. L., and Erickson, W. P.

2002. Resource selection by animals: statistical design and analysis of field studies.

Kluwar Academic Publishers, Dordrecht, The Netherlands. pp. 29.

Martino, J. A., Poulin, R. G., Parker, D. L., and Somers, C. M. 2012. Habitat selection by grassland snakes at northern range limits: implications for conservation. J. Wildl.

Manage. 76(4): 756-767.

Moore, J. A. and Gillingham, J. C. 2006. Spatial ecology and multi-scale habitat selection by a threatened rattlesnake: the eastern massasauga (Sistrurus catenatus catenatus).

Copeia, 2006(4): 742-751.

Moriarty, J. J. and Linck, M. 1997. Reintroduction of bullsnakes into a recreated prairie.

In: Moriarty, J. J. and Jones, D. 1997. Minnesota’s amphibians and reptiles: their conservation and status. Serpent’s Tale Natural History Book Distributors, Lanesboro,

Minnesota. pp. 43-54.

Parker, J. M. and Anderson, S. H. 2007. Ecology and behavior of the midget faded rattlesnake (Crotalus oreganus concolor) in Wyoming. J. Herpetol. 41(1): 41-51.

Parks Canada Agency. 2010. Recovery strategy for eastern yellow-bellied racer (Coluber constrictor flaviventris) in Canada. Species at Risk Act Recovery Strategy Series. Parks

Canada Agency. Ottawa. p. 12.

Robel, R. J., Briggs, J. N., Dayton, A. D., and Hulbert, L. C. 1970. Relationships between visual obstruction measurements and weight of grassland vegetation. J. Range. Manage.

23(4): 295-297.

Rodriguez-Robles, J. A. 2003. Home ranges of gopher snakes (Pituophis catenifer,

Colubridae) in central California. Copeia, 2003(2): 391-396.

Row, J. R. and Blouin-Demers, G. 2006. Kernels are not accurate estimators of home- range size for herpetofauna. Copeia, 2006(4): 797-802.

Schaefer, W. H. 1934. Diagnosis of sex in snakes. Copeia, 4: 181.

Smith, R. M., Spotila, J. R., and Bien, W. F. 2015. Spatial ecology of northern pinesnakes

(Pituophis m. melanoleucus) in disturbed and undisturbed habitats in the New Jersey pine barrens. Herpetologica, 71(1): 19-25.

Sperry, J. H. and Weatherhead, P. J. 2009. Does prey availability determine seasonal patterns of habitat selection in Texas ratsnakes? J. Herpetol. 43(1): 55-64.

Thomas, D. L. and Taylor, E. J. 2006. Study designs and tests for comparing resource use and availability II. J. Wildl. Manage. 70(2): 324-336.

Webb, J. K. and Shine, R. 1998. Using thermal ecology to predict retreat-site selection by an endangered snake species. Biol. Conserv. 86: 233-242.

90: 304-312.

Wisler, C., Hofer, U., and Arlettaz, R. 2008. Snakes and monocultures: habitat selection and movements of female grass snakes (Natrix natrix l.) in an agricultural landscape. J.

Herpetol. 42(2): 337-346.

Wright, J. D. 2008. Bullsnake, Pituophis catenifer sayi, nesting biology in Alberta. Can.

Field Nat. 122(2): 138-141.

Zappalorti, R., Burger, J., and Peterson, F. 2015. Home range size and distance traveled from hibernacula in northern pinesnakes in the New Jersey pine barrens. Herpetologica,

71(1): 26-36.

Chapter 3: Synthesis

Conservation and management implications

My study is one of the first to address bullsnake (Pituophis catenifer sayi) spatial ecology at the northern range limits of this species in Canada. I found that space use is highly variable among bullsnakes, with a 58 fold difference between the minimum and maximum home range estimates among individuals in the Saskatchewan populations.

The level of variation measured here exceeds that previously observed for bullsnakes in the United States (Moriarty and Linck 1997; Fitch 1999; Rodriguez-Robles 2003; Kapfer et al. 2008). This study is also one of the first to measure the distance bullsnakes travel away from overwintering den sites; data that may be relevant to many populations in the

United States that rely on den sites for overwinter survival. I found that bullsnakes can move short distances from den sites (as little as 352 m) or may undergo long distance migrations between overwintering and summering habitats (up to 3.9 km). Space use and movement data collected here indicate that bullsnakes may have greater space requirements than previously thought. Though management plans are not currently in place for bullsnakes in Canada, extrapolating space use patterns from one population to another does not appear to be appropriate, as I found spatial ecology to differ among populations.

Population-specific conservation and management strategies are recommended for snakes, as space use and movement patterns vary intraspecifically among snake populations. In addition to bullsnakes, variation in spatial ecology has been documented for other snake species, including the eastern massasauga rattlesnake (Sistrurus catenatus; Marshall et al. 2006; Moore and Gillingham 2006; Bailey et al. 2012), prairie rattlesnake (Crotalus viridis; Jorgensen 2008; Gardiner et al. 2013; Shipley et al. 2013;

Bauder et al. 2015), and midget-faded rattlesnake (Crotalus oreganus concolor; Parker

and Anderson 2007; Gomez et al. 2015). Issues with making generalizations among intraspecific populations have previously been examined for two snake species in

Canada: the eastern yellow-bellied racer (Coluber constrictor flaviventris) and Great

Basin gophersnake (Pituophis catenifer deserticola). In Canada, racer den sites on federal lands are surrounded by 500-m conservation buffers of critical habitat. However, 500 m buffers are not sufficient for racers in Saskatchewan, which can travel up to 5 km from den sites (Martino et al. 2012; Gardiner et al. 2013). Great Basin gophersnake den sites are surrounded by 200 to 300 ha wildlife habitat areas in British Columbia; however, many gophersnakes travel outside of these areas and their space use patterns do not match the circular shape of the protected area (Williams et al. 2012). Rather than make generalizations, conservation and management plans should attempt to use site-specific data to inform conservation strategies for populations of concern. However, it may not always be feasible to individually assess each population of a species in terms of their space requirements. As such, I recommend implementing protective buffer zones surrounding den sites that encompass the maximum distance a snake of that species has been documented to travel. For example, this study demonstrated (in addition to Martino et al. 2012 and Gardiner et al. 2013) that bullsnakes may travel up to 4 km from den sites.

As such, a 4 km protective buffer zone will likely encompass the summer and overwintering habitat of most bullsnakes within a population based on current knowledge. Furthermore, researchers may attempt to supplement site-specific data with data from other regions or related species, although this should be minimized (Kapfer et al 2008).

The observed variation in space use is likely the result of a similar variation in habitat and resource availability among populations (Moore and Gillingham 2006; Kapfer

et al. 2008). As such, habitat selection also varies among intraspecific snake populations

(Carfagno and Weatherhead 2006; Bauder et al. 2015; Gomez et al. 2015). Bullsnakes in

Saskatchewan, as well as across their geographic range, appear to be quite generalist in their third order habitat selection. Bullsnakes in the Frenchman River Valley selected for lowland pasture, hills, and roads (Martino et al. 2012; Gardiner et al. 2013), snakes in the

Big Muddy Valley used native pasture based on availability and selected for farmyards, while snakes in the South Saskatchewan River Valley used native prairie based on availability, but selected for a wider diversity of modified-habitats including beach, tame fields, mowed areas, and roads. Overall, bullsnakes in Saskatchewan appear to use native habitats based on their availability. Human modified habitats, however, were used 2 times more than expected in the Big Muddy and South Saskatchewan River Valleys, compared to 3 times less than expected in the Frenchman River Valley. In the United

States, bullsnakes select primarily for prairie habitat (Moriarty and Linck 1997;

Rodriguez-Robles 2003) as well as open bluff areas (Kapfer et al. 2008). The only observed similarity between bullsnake populations in terms of habitat selection appears to be that this species selects primarily for open habitats with minimal tree cover. However, as with space use patterns, it is not advisable to make generalizations about snake habitat selection among populations. Habitats available in one area may not be available in others and relative habitat availability will differ among landscapes and thus, populations.

Regardless of third order habitat selection, retreat sites were important for bullsnakes in Saskatchewan at a fine spatial scale. Bullsnakes were most frequently found within 0 to 1 m of a retreat site (also observed by Martino et al. 2012). Such sites provide cover for thermoregulation and the opportunity for prey capture and predator avoidance.

Selection for areas in close proximity to retreat sites has been observed for other snake

species (Harvey and Weatherhead 2006b in eastern massasauga rattlesnakes; Halstead et al. 2009 in coachwhips; Martino et al. 2012 in eastern yellow-bellied racers; Gardiner et al. 2015 in prairie rattlesnakes), demonstrating the importance of retreat sites for many snake species in general.

Retreat sites used by bullsnakes were both natural and man-made. The introduction of human-modified habitats and thus, man-made features (including retreat sites) is potentially beneficial to snakes. For example, snakes in the South Saskatchewan

River Valley were found under walkways, in parking lots under cement blocks, and under buildings. Snakes using these retreat sites occupied these sites over longer periods of time, potentially indicating that they were suitable for thermoregulation and for meeting other snake needs. Previous research has found that snakes use artificial structures on the landscape, such as buildings and wells, as overwintering den sites (Woodbury and Parker

1956; Costanzo 1986; Burger et al. 1988). As such, man-made retreat sites may be beneficial to snakes. However, use of these habitats may also be costly for snakes. The use of these man-made habitat features in areas of high human activity could potentially increase risk of mortality via snake-vehicle collisions or human persecution. As such, when assessing how the introduction of modified habitats will affect snake populations, researchers should consider the potential threats to snakes making use of these introduced habitat features. In addition, the attraction of bullsnakes to human-modified habitats may result in a source-sink population dynamic, where modified habitats do not benefit snake survival or fitness. Source populations may in fact occupy native habitats, which benefit snake survival, reproduction, and fitness.

Mammal-created burrows appear to be an important habitat feature for bullsnakes.

Natural retreat sites were comprised of small mammal burrows, and burrow systems were

also used as overwintering den sites by bullsnakes in Saskatchewan. Information regarding small mammal populations should therefore be used to inform and develop management strategies for bullsnakes. Critical habitat is defined as areas occupied by a species containing physical and biological features that are essential to the survival and recovery of a species (Environment Canada 2009). Although this study did not identify critical bullsnake habitat at the landscape scale, it did identify retreat sites to be an important habitat feature for bullsnakes that should be considered when developing conservation and management strategies.

In areas where natural refuge or den sites are limited, artificial den and refuge sites may be constructed. Previous research has demonstrated it is possible to construct artificial structures that meet the overwintering needs of snakes, while also meeting other habitat requirements for thermoregulation and nesting (Gillingham and Carpenter 1978;

Zappalorti and Reinert 1994; Ernst 2003). In cases where natural den sites are under threat from human development, the construction of successful artificial dens would be an important tool to reduce the loss of overwintering habitat for snake populations.

Future research

Future research should focus on identifying the resources driving differences in space use among bullsnake populations in Saskatchewan. This study demonstrated that the spatial association among seasonal habitats is an important determinant of bullsnake space use. However, the resources driving the seasonal migration of snakes away from den sites remain to be addressed. As mentioned in Chapter 2, prey and refuge sites are likely the determining resources of bullsnake summer space use. Measuring the differences in the availability and abundance of these resources across landscapes will help to explain the observed differences in space use among populations. Data collected

here are also potentially applicable to other populations of the species in terms of understanding the driving factors of snake summer space use. Future studies may also aim to link habitat selection to a measure of individual snake fitness (similar to Blouin-

Demers and Weatherhead 2008 using locomotion speed as a proxy for fitness), to measure the costs or benefits of snakes selecting for different habitats.

As retreat sites are an important habitat feature for bullsnakes, future research should also aim to identify the specific physical and thermal characteristics of the retreat sites selected by bullsnakes. There were many occasions during snake tracking in

Saskatchewan on which two bullsnakes used the same burrow simultaneously or snakes returned to the same burrow (or man-made retreat site) on multiple occasions over the summer. Identifying the specific characteristics that bullsnakes are selecting for in their retreat sites will contribute to an understanding of bullsnake ecological needs and fine- scale habitat requirements. This is especially important as few studies have previously identified the distinct thermal and physical characteristics of snake refuge sites (Huey et al. 1989; Webb and Shine 1998; Croak et al. 2008).

In addition, future studies should aim to examine the thermal and physical properties of snake hibernacula, as overwintering den sites are an important limiting resource for snakes at northern range limits. The external physical characteristics of these sites may be quantified by measuring various structural and vegetation parameters, similar to Burger et al. (1988), Prior and Weatherhead (1996), and Harvey and

Weatherhead (2006a). Few studies have examined the internal characteristics of den sites, due to the disruptive nature of den excavation (Schroder 1950; Burger et al. 1988;

Rudolph et al. 2007). Instead, I recommend using ground-penetrating radar to examine the subsurface structure of hibernacula, as this would be less invasive. Thermal regimes

within den sites should also be measured by quantifying the thermal gradient from den site entrance to within the hibernacula itself. This would allow for an increased understanding of the changes in body temperature snakes undergo during the overwintering period. Measuring the physical and thermal characteristics of various den sites would also help to determine if certain den sites are of higher quality in terms of increasing snake overwinter survival.

A significant knowledge gap currently remains to be addressed for bullsnakes in terms of their nesting ecology (Wright 2008). Where nesting sites may be identified, future research projects should aim to identify nesting habitats by tracking female bullsnakes, using radio-telemetry, from the start of the active season to the nesting season. It will also be important to examine differences in thermal ecology, habitat selection, and space use among gravid and non-gravid females, as well as among male and female bullsnakes. Certain habitat features or habitat characteristics may be important to different age classes, males versus females, or gravid versus non-gravid snakes within a population. Such differences need to be addressed to accurately inform any conservation and management decisions for these snakes.

Finally, the sample size of the current study was relatively small, specifically in the Big Muddy Valley. In the Big Muddy Valley 7 snakes were tracked, whereas 14 snakes were tracked in the South Saskatchewan and Frenchman River Valleys. The number of den sites sampled was also relatively small across all river valleys. As such, increasing sample size (for both individual snakes and overwintering den sites) will be critical to further understanding the extent to which habitat and space use patterns may vary among bullsnake populations. Multi-year studies are particularly beneficial, as they identify the extent to which space and habitat use may vary among years. Such studies

will allow for the examination of bullsnake den site fidelity over years within the different river valleys.

References

Bailey, R. L., Campa III, H., Bissell, K. M., and Harrison, T. M. 2012. Resource selection by the Eastern massasauga rattlesnake on managed land in Southwestern Michigan. J.

Wildl. Manage. 76(2): 414-421.

Burger, J., Zappalorti, R. T., Gochfeld, M., Boarman, W. I., Caffrey, M., Doig, V.,

Garber, S. D., Lauro, B., Mikovsky, M., Safina, C., and Saliva, J. 1988. Hibernacula and summer den sites of pine snakes (Pituophis melanoleucus) in the New Jersey pine barrens. J. Herpetol. 22(4): 425-433.

Carfagno, G. L. F. and Weatherhead, P. J. 2006. Intraspecific and interspecific variation in use of forest-edge habitat by snakes. Can. J. Zool. 84: 1440-1452.

Costanzo, J. P. 1986. Influences of hibernaculum microenvironment on the winter life history of the garter snake (Thamnophis sirtalis). Ohio J. Sci. 86(5): 199-204.

Croak, B. M., Pike, D. A., Webb, J. K., and Shine, R. 2008. Three-dimensional crevice structure affects retreat site selection by reptiles. Anim. Behav. 76: 1875-1884.

Environment Canada. 2009. Species at Risk Act policies. [electronic resource].

Ernst, R. 2003 Lethbridge prairie rattlesnake conservation project: 2002/2003 progress report. Alberta species at risk report no. 64.

Fitch, H. S. 1999. A Kansas snake community: composition and changes over 50 years.

Krieger Publishing Co., Malabar, Florida. pp. 105.

Gardiner, L. E., Somers, C. M., Martino, J. A., Parker, D. L., and Poulin, R. G. 2013.

Balancing the dumbbell: summer habitats need protection in addition to winter dens for northern snake communities. J. Wildl. Manage. 77(5): 975-982.

Gillingham, J. C. and Carpenter, C. C. 1978. Snake hibernation: construction of and observations on a man-made hibernaculum (Reptilia, Serpentes). J. Herpetol. 12(4): 495-

498.

Gomez, L., Larsen, K. W., and Gregory, P. T. 2015. Contrasting patterns of migration and habitat use in neighboring rattlesnake populations. J. Herpetol. 49(3): 371-376.

Halstead, B. J., Mushinsky, H. R., and McCoy, E. D. 2009. Masticophis flagellum selects for Florida scrub habitat at multiple spatial scales. Herpetologica, 65(3): 268-279.

Harvey, D. S. and Weatherhead, P. J. 2006a. Hibernation site selection by eastern massasauga rattlesnakes (Sistrurus catenatus catenatus) near their northern range limit. J.

Herpetol. 40(1): 66-73.

Harvey, D. S. and Weatherhead, P. J. 2006b. A test of the hierarchical model of habitat selection using eastern massasauga rattlesnakes (Sistrurus c. catenatus). Biol. Conserv.

130: 206-216.

Huey, R. B., Peterson, C. R., Arnold, S. J., and Porter, W. P. 1989. Hot rocks and not-so- hot rocks: retreat site selection by garter snakes and its thermal consequences. Ecology,

70(4): 931-944.

M. D., and Bush, S. P. The biology of rattlesnakes. Loma Linda University Press, Loma

Linda, California. pp. 303-316.

Kapfer, J. M., Coggins, J. R., and Hay, R. 2008. Spatial ecology and habitat selection of bullsnakes (Pituophis catenifer sayi) at the northern periphery of their geographic range.

Copeia, 2008(4): 815-826.

Martino, J. A., Poulin, R. G., Parker, D. L., and Somers, C. M. 2012. Habitat selection by

grassland snakes at northern range limits: implications for conservation. J. Wildl.

Manage. 76(4): 756-767.

Marshall, J. C. Jr., Manning, J. V., and Kingsbury, B. A. 2006. Movement and macrohabitat selection of the eastern massasauga in a fen habitat. Herpetologica. 62(2):

141-150.

Moore, J. A. and Gillingham, J. C. 2006. Spatial ecology and multi-scale habitat selection by a threatened rattlesnake: the eastern massasauga (Sistrurus catenatus catenatus).

Copeia, 2006(4): 742-751.

Moriarty, J. J. and Linck, M. 1997. Reintroduction of bullsnakes into a recreated prairie.

In: Moriarty, J. J. and Jones, D. 1997. Minnesota’s amphibians and reptiles: their conservation and status. Serpent’s Tale Natural History Book Distributors, Lanesboro,

Minnesota. pp.43-54.

Parker, J. M. and Anderson, S. H. 2007. Ecology and behavior of the midget faded rattlesnake (Crotalus oreganus concolor) in Wyoming. J. Herpetol. 41(1): 41-51.

Prior, K. A. and Weatherhead, P. J. 1996. Habitat features of black rat snake hibernacula in Ontario. J. Herpetol. 30(2): 211-218.

Rodriguez-Robles, J. A. 2003. Home ranges of gopher snakes (Pituophis catenifer,

Colubridae) in central California. Copeia, 2003(2): 391-396.

Rudolph, D. C., Schaefer, R. R., Burgdorf, S. J., Duran, M., and Conner, R. N. 2007. Pine snake (Pituophis ruthveni and Pituophis melanoleucus lodingi) hibernacula. J. Herpetol.

41(4): 560-565.

Schroder, R. C. 1950. Hibernation of blue racers and bullsnakes in western Illinois. Nat.

Hist. Miscell. 75: 1-2.

Webb, J. K. and Shine, R. 1998. Using thermal ecology to predict retreat-site selection by an endangered snake species. Biol. Conserv. 86: 233-242.

90: 304-312.

Woodbury, A. M. and Parker, D. D. 1956. A snake den in cedar mountains and notes on snakes and parasitic mites. Herpetologica. 12(4): 261-268.

Wright, J, D. 2008. Bullsnake, Pituophis catenifer sayi, nesting biology in Alberta. Can.

Field Nat. 122(2): 138-141.

Zappalorti, R. T. and Reinert, H. K. 1994. Artificial refugia as a habitat-improvement strategy for snake conservation. Contrib. Herpetol. 11: 369-375.